Blog

StarTech SATA Duplicator/Dock

Fri, 3 Feb 2012 09:09:01 -0600

Usually when we evaluate products the tests can be a little contrived. We often have to copy a test data set a number of times or come up with a testing scenario that’s similar to what we think a user would encounter. An interesting thing happened when I got ready to try out the StarTech SATA Duplicator recently - I really needed to use it.

We had just gotten some SSDs in and I was preparing to put one into a MacBook Pro that I use. I simply pulled the existing 2.5” hard drive out of the notebook and connected it to the StarTech Duplicator, along with the SSD. Here are the particulars:

Source: Hitachi 5K500 160GB 2.5" 5400 RPM SATA drive
Destination: OCZ Deneva 2, 240GB 2.5” eMLC SATA drive

1. Plugged the StarTech device into the power adaptor
2. Attached both drives
3. Hit the power button on the device
4. Pushed "Mode" to switch from “Dock” to “Duplicate”
5. Hit "Start"

That was it. As my teenage son would say it was "stupid easy". The Duplicator has four LEDs that show you the status of your copy job and I was done in 51 minutes and 10 seconds - copying 112GB of data. The spec sheet said the StarTech Duplicator can copy at up to 72MB/s. According to my math, I got half of that, which doesn’t seem unusual given the variability of storage devices. When the copy was done I plugged the USB cable into another Mac, switched the Duplicator to “Dock” mode and both drives showed up in Finder - both with 112GB of data.

I should point out that the StarTech Duplicator/Dock also works with essentially all Windows OSs and Linux. I just happened to be using a Mac. In Dock mode it connects one or two drive via USB 2.0, or eSATA with the proper host controller and Port Multiplier (in order to see two drives with eSATA). The device can copy up to a 3TB capacity SATA drive, 2.5” or 3.5” form factor - at least that’s the largest capacity drive they’ve tested to date, but plan to update that to 4TB drives in the near future.

For laptop and desktop support people in a corporate IT department this product would be invaluable, since it lets you make an exact, sector-for-sector copy of a hard drive - outside of the computer. As a USB hard drive dock, it would make the job of managing the accumulated inventory of orphaned disk drives easier, not to mention drive upgrades and replacements. When somebody drops their laptop, this little duplicator could save a lot of steps getting them back up and running on a new computer. That got me to thinking.

I travel quite a bit, especially during trade show seasons in the spring and fall each year. As a writer my business is on my laptop, literally. Of course I back it up numerous ways but when I’m out of town or away from the office, I’m a little vulnerable. If I drop the laptop for example, things can get tense.

If I have everything backed up to a USB drive or the cloud I could be OK. There are certainly numerous other ways to prevent data loss and get around some of the hassle, but what about this for a solution:

Pull the drive out of the dead laptop - it’s most likely fine since the laptop chassis took the brunt of the shock of being dropped. Plug it into the StarTech Dock, plug the Dock into the USB slot on another computer and you’re in business - nothing to copy. If you have a presentation to show or a series of critical files, they’re all there, looking exactly as they did on the old computer. This could allow the use of anyone’s computer and save a lot scrambling before an important meeting.

Or, pull out the drive from a new laptop (same make and model, etc) and copy the old hard drive to the new one. You don’t need another computer to do this copy operation, it could happen in a hotel room. Then reinstall it in the new laptop and you’re done. For Mac users this is especially attractive since there are lots of Apple stores around and fewer models to confuse the issue.

Again, there are a lot of ways to provide protection for data and get you up and running if you have a problem. I’m just saying this is one of the easiest and as a device that’s small enough to fit into your hand, it’s certainly portable enough to take in your computer bag. In addition to the time it can save in the office or lab, for about $75 the StarTech SATA Duplicator/Dock is also cheap insurance on the road.

Recent Briefing Notes
Third Generation All-Flash Storage Systems for the Enterprise
Cutting The Price on 1 Million IOPS
Taming the Unstructured Data Beast

StarTech is not a client of Storage Switzerland

previous entry: “Third Generation All-Flash Storage Systems for the Enterprise”

Nimbus E-Class

Tue, 31 Jan 2012 10:20:57 -0600

Last August Nimbus released the second generation of their S-Class, all flash storage system. These are scalable, rack-based storage systems comprised of 2U blades, each providing up to 10TB of flash capacity or 250TB of total capacity per system. Now Nimbus is announcing its third generation of a pure flash storage system, the E-Class, an enterprise version this ‘sustainable storage’ technology. The E-Class can expand up to 500TB per system in their non-blocking internal fabric architecture, producing up to 800K IOPS and 8GB per second of throughput.

The E-Class features an enterprise-grade, completely redundant design, with no single points of failure. This includes redundant controllers with automatic failover, non-disruptive software upgrades and non-disruptive capacity and file system expansion. Other high availability features include active-active I/O modules plus hot swap flash modules, power supplies and fans. Nimbus’s eMLC NAND flash systems support RAID 5, 6 and 10 with 28% reserve flash capacity to improve performance, durability and reliability.

In addition, the E-Class makes big strides in the area of efficiency, with power usage as low as 5 Watts per TB of capacity. This system also leads the industry in space efficiency, providing up to 10TB per U or 440TB per rack. This translates into roughly an 80% lower cooling requirement than a comparable disk array system.

The Nimbus E-Class provides unified storage, SAN and NAS in a single file system, and multi-protocol support - GbE, 10GbE, Fibre Channel and Infiniband. The HALO operating system provides the storage services enterprise users have come to expect. Storage virtualization, thin-provisioning, deduplication, snapshots, mirroring and replication are included in all Nimbus storage systems.

Storage Swiss Take

Flash-only systems are now a category unto themselves, with four competitors in the space. Nimbus’s quickly maturing product line offers the high performance you would expect in a flash-only system but adds space and power saving features that for many data centers are just as important.

Recent Briefing Notes
Advancing the State of Solid State Caching
What to Look for in Server Based SSD
The Year of the SSD Cache
What is Server Based Solid State Caching?

Nimbus is not a client of Storage Switzerland

previous entry: “Catching Issues Before they Become Career Limiting”

Catching Issues Before they Become Career Limiting

Mon, 30 Jan 2012 13:13:22 -0600

Virtualization administrators have a big job, they’re tasked with keeping the VM environment running. This can include a number of areas (but isn’t limited to): fixing problems before they impact users, supporting the expected growth of virtual machines, maintaining performance levels during that growth and keeping infrastructure cost in line the whole time. Unfortunately for IT professionals, balancing VMs’ resource needs and resource availability with the usual management utilities can be like trying to control a moving train - from the platform.

For the administrators who are out in front of this freight train, a set of comprehensive performance monitoring tools isn’t a luxury, it’s a requirement. Making poor decisions, or no decisions at all, can be career limiting. These administrators need intelligence, not raw data; recommendations and suggested courses of action not just more information. They need effective tools that can see the entire environment but not overwhelm them with so much information that it obscures the critical metrics.

What

NetApp’s OnCommand Insight Balance is a performance and capacity management software solution that enables administrators to optimize resource usage and assure performance of critical VMs. It collects raw data from individual elements in the environment - servers, storage systems, network components and VMs - then automatically maps and analyzes the virtual infrastructure from end to end, without agents. Then it takes this raw data and applies powerful analytics to produce the intelligence needed to understand how resources and application workloads interact and to generate actionable information.

OnCommand Insight Balance produces detailed metrics to maximize performance, speed troubleshooting and manage utilization, from end to end. It enables IT managers and administrators to manage their infrastructure from virtual machines and physical servers, to storage arrays and individual spindles.

Who cares about Performance?

Virtualization is no longer a project being run on test and development servers or restricted to applications that no one really sees. VMware projects are now into later phases of implementation, which means production applications are going onto VMs. These are visible all the way up to the executive offices. You can’t hide poor performance of a critical application; people care, somebody will come looking. Virtualization administrators have to get on top of their performance issues. The question is: “where to start?”.

IT Infrastructure Key Performance Indicators

OnCommand Insight Balance provides information on critical aspects of a shared infrastructure that can help you pinpoint problems and often prevent them. Balance creates Key Performance Indicators for the IT department to assess and manage their shared infrastructure against, including Infrastructure Response Time, Performance Index and Disk Utilization.

The Performance Index (PI) is a value calculated for each host based on CPU, Memory and Storage, the three resources that most impact performance. A PI value of 100 is an optimal point or a baseline, indicating a host is running at near-perfect efficiency. A host with values below 100 indicates it’s underutilized and can support more workloads. A PI value above 100 means the host is overworked and should have some VMs removed or more resources added, such as more memory or better-performing storage.

Using the Performance Index enables the virtualization team to know where best to put new VMs in advance, by seeing which hosts are below their optimal loading value. They can also see which hosts are overworked and represent a problem in the making, even before current applications start slowing down and complaints come in from users.

Infrastructure Response Time (IRT) is a metric that tells how long it takes for a workload to get finished, which is really the operable metric. (see graphic below) Latency, not absolute statistics like IOPS or capacity used, is what indicates a problem, or a potential problem. IRT shows just that, how well an application workload is being serviced by the infrastructure. When IRT spikes, that means an application is waiting for resources and a problem is in the making.

Disk Utilization is a predictive analytic that can show if the storage system can handle more workloads. This can help keep you from overloading the storage infrastructure and killing performance for existing VMs. Together with the Performance Index it can help keep the infrastructure balanced and optimized for the available resources.

Stuff happens

Dealing with the issues that will come up means getting notification of problems or hopefully, potential problems, so they can be dispatched quickly. OnCommand Insight Balance displays storage arrays, VMs and virtual hosts that are in trouble or at risk (see Dashboard graphic below).

Users can also see the “victims” involved, the server instances which are impacted, so they can tell how critical the problem is. As the disk contention analysis screenshot below shows, OnCommand Insight Balance identifies the root cause, or the “bully” server which started the problem.

Who Cares about Optimization?

Virtualization was typically sold to management as a way to get more ‘bang for the IT buck’. This is a message that’s been embraced by CIOs and IT Directors who are trying to do more with less. The only problem is they’re not the ones making sure that VMware environment is actually performing as expected, and producing the ROIs that were promised. That’s the job of the server and storage teams. Resource optimization is something that ends up being important to many people.

Getting the most out of the resources supporting this infrastructure requires a thorough understanding, a familiarity with how storage, CPU cycles and memory capacity are balanced. It’s a real-time, ongoing tuning process, not a rote exercise. A dynamic virtual server infrastructure can change almost overnight, making it difficult to stay on top of those changes and keep the balance. OnCommand Insight Balance is designed to do just that.

Would you rather use a GPS or a Compass?

Optimizing resources means pushing a little closer to the limit, which can make people nervous; and it should if they don’t have the information they need. Pulling back can lower utilization and increase costs, which may not hit ROI targets that management is expecting. Monitoring tools that are focused on individual elements in the environment can present a mountain of data, but don’t always produce the actionable intelligence that’s needed.

OnCommand Insight Balance takes the abundance of available data and applies powerful analytics to generate this intelligence, providing the confidence to push for real optimization. An analogy could be navigating a car with a GPS unit instead of a compass. The administrators driving this car will be much more willing to put the throttle of resource utilization down when they have accurate, real-time intelligence about where they are and where they’re going.

IT problems aren’t always totally clear cut and the cause of an issue often overlaps multiple groups, like the storage, networking and server teams. In the abstract world of virtualized resources, often nobody really has enough information, which can lead to the ‘blame game’. So, in addition to making the right decisions, virtualization administrators have to defend them as well; like the math teacher used to say: “show your work”. OnCommand Insight Balance can produce the data needed to justify decisions and identify where the problem does or doesn’t lie.

Storage Swiss Take

For IT administrators keeping the virtual server infrastructure up and running at an optimum level is the prime directive. Their careers depend on preventing performance issues when they can, fixing problems when they have to and being able to justify their actions when they’re asked. And all of this must be done in a changing environment, without wasting resources. NetApp OnCommand Insight Balance provides the right information at the right time to satisfy this prime directive so that issues don’t become career limiting.

Related Content
Compellent Storage Center 6.0 - 64bit, Enterprise Ready
Dell-Compellent Upgrades To Storage Center 6.0
Dell Brings Advanced Backup & SR to the SMB Market
iSCSI 2.0 - Using Data Center Bridging to Enhance iSCSI

Dell is a client of Storage Switzerland

previous entry: “Compellent Storage 6.0 - 64bit, Enterprise Ready”

CTERA - Test Drive Update

Fri, 27 Jan 2012 11:22:49 -0600

Our testing of the CTERA Cloud Attached Storage appliance continues to roll along. We continue to be impressed with the units versatility as a NAS, SAN, backup target and of course as a gateway to cloud storage. We have now tested its NAS functionality and backup capability fully as well as its replication to the cloud and all have worked as advertised.

As you can see in the below video the first order of business was to make sure that we showed the completion of the replication of our backup jobs from our last test to the cloud. This worked flawlessly. The backup’s have now been running for over a month on our Windows 2008 servers with no problems encountered thus far.

From a NAS perspective we have used the CTERA appliance to serve up CIFS(Windows), NFS (UNIX/Linux) and AFP (Apple) volumes and then have specific folders of our choosing on those shares replicated to the cloud. This is what makes the CTERA Cloud Attached Storage appliance so unique, the multiple ways it provides you to move or copy data to a cloud storage host. There are options to backup, synchronize or directly store data on the local appliance all of which then can be replicated to cloud based storage.

CTERA provides an excellent solution for businesses looking to start leveraging the cloud but still have a copy of their data locally. It also is an ideal platform for Managed Service Providers to offer to their customers allowing them to get into the cloud storage market place. In either case the unit deserves strong consideration.

Previous Blogs

Cloud Attached Storage May be the Answer
Using Cloud Attached Storage for Backups
All in One Cloud Solution for Service Providers
Do it All Cloud Attached Storage C400 Test Drive
Gateways Help Cloud Service Providers

CTERA is a client of Storage Switzerland

Previous Entry: “Storage Center 6.0 - 64bit, Enterprise Ready”

Compellent Storage Center 6.0 - 64bit, Enterprise Ready

Thu, 26 Jan 2012 08:53:01 -0600

Dell Compellent recently announced the upgrade of their storage virtualization software known as Storage Center to version 6.0. While this release has some external improvements, including increased support of VMware’s VAAI API set, the importance of this release is its move to a 64-bit operating system. With Storage Center 6.0 Compellent’s systems are now well able to compete with virtually any enterprise-class storage system.

The Dell Compellent systems have been successful storage options for mid-range and larger data centers, as well as for project-specific deployments in full-scale, enterprise environments. The lack of a 64-bit operating system though has kept Compellent from competing and being the single storage consolidation play in these very large and enterprise data centers. With version 6.0 and the access to new Intel-powered storage server hardware by leveraging Dell’s relationship with Intel, there is no reason to believe that Compellent won’t be able to provide a single storage system that will meet all the demands of the largest of data centers.

Storage systems all have to maintain information about the data they are storing. How Compellent Storage Systems manages this ‘data about data’ (called metadata) is one of the key changes within Storage Center 6.0. This metadata can be simply the location of actual data, or can be more complex information. Examples are data about the mapping of snapshots, the allocation of storage to meet the provisioning demands and the tracking of blocks to position them on different tiers of storage. The size of memory needed to hold the metadata tables, where all this information is stored, has increased because of the demand for higher capacity systems but also because of the demand for more features and automation in those storage systems.

The transition from a 32-bit to a 64-bit operating system increases the potential memory address space from 4 GB to 16 Exabytes (16,000 PB) to support this larger table space. This shift to a 64-bit architecture represents a significant upgrade for metadata operations like those described above. It also paves the way for new capabilities, like deduplication for example, that the constraints of a 32-bit operating system simply would not allow.

64-bit performance and scale

Increasing the potential amount of address space for metadata allows for larger systems with greater spindle counts and higher capacity disk drives. As mentioned above each segment of data written to the storage system has to be tracked via metadata. That data can create an expanding amount of metadata thanks to snapshots and other storage services, making a single data block contribute significantly to the size of the total storage capacity being managed.

Adding more spindles doesn’t just help solve the capacity problem but also helps with overall system performance. Since virtualized storage systems typically use a higher number of drives shared across multiple volumes there is the potential for configuring high drive count volumes, leading to better overall system performance without significantly adding to storage administration time.

The increase in the available memory address space for metadata provides four key benefits. These should be of value to existing Dell Compellent customers and be appealing to the IT manager as well, prompting them to consider Dell Compellent for their next storage refresh.

Platform for growth

The 64-bit operating system is also the required first step in integrating the other components of Dell-owned IP that made the Compellent acquisition so interesting. With this platform in place Compellent is well on its way to using higher-end and greater performing storage server platforms as well as integrating key Dell technologies like Ocarina’s deduplication, ExaNet’s fluid data file system and RNA’s host level caching.

With the 64-bit Storage Center we expect the feature richness of the Compellent platform to continue and even accelerate as these complementary components of Dell IP are integrated. When this happens, Dell will be able to provide one of the most scalable and complete storage platforms in the industry.

Good news for existing customers

Existing customers will be able to access the 64-bit OS via a download at no charge if they have an active service contract. For customers already experiencing a problem with the addressable memory space they can buy a memory upgrade kit that takes their existing storage controllers from 6 GB to 12 GB of memory per controller. This should immediately resolve those issues and allow them to expand their existing systems further. The upgrade is non-disruptive for those customers using dual, clustered controllers.

Good news for prospective customers

For data centers that are looking to move to a new platform for a specific project or are looking for a storage system that they can consolidate onto, the 64-bit Storage Center makes the Compellent product offering worthy of consideration. These data centers are now no longer limited to the big three and their pricing structures.

Storage Swiss Take

The enterprise storage system choice is no longer limited to the big three and their somewhat complex pricing models. Dell Compellent with their new 64-bit code brings enterprise class scalability and a more simplified pricing strategy on a storage system that has been one of the leaders in storage feature innovation. The combination makes for a strong consideration for companies looking to refresh their storage strategy.

Related Content
Dell-Compellent Upgrades To Storage Center 6.0
Dell Brings Advanced Backup & SR to the SMB Market
iSCSI 2.0 - Using Data Center Bridging to Enhance iSCSI

Dell is a client of Storage Switzerland

previous entry: “Test Driving Drobo DR”

Test Driving Drobo DR

Mon, 23 Jan 2012 08:29:34 -0600

In our last video we showed how to set up and configure Drobo's new DroboSync capability on our Drobo B800fs. As with all things Drobo it just worked. DroboSync was incredibly simple and easy to configure and it has proven to be very reliable in operation. We have been replicating data to our DR Drobo B800fs for months now.

Reliable operation is important because disaster recovery is called on to work when the pressure is high and if a simple and easy operation means an unreliable recovery then you might as well not have DR in the first place. In this test drive we put DroboSync to the test and I'm happy to report that the actual recovery is even easier than the initial setup.

Our first DR test for DroboSync was a simple file recovery from the DR box. You might need to do this for a couple of reasons, either your primary Drobo B800fs has failed or you could use it to recover a corrupted file. The later use case is an interesting use of the replication capabilities. The replication interval can be set in a range of 2-24 hours. As long as you catch the corruption in that time you can copy the prior version of the file back to its original location.

As I said earlier, operation of the replication function has been flawless. We have been syncing from our primary Drobo B800fs for several months now with no issues. Replication happens in the background automatically, we don't have to think about it or monitor it. In the SMB this is very important since the "IT Guy" often has many other tasks going on and won't have the time to baby sit a DR job. For example we have lost power several times and our WAN connection went down a couple of times, each time DroboSync found its partner and resumed replication.

As the video below shows we were able to "accidentally" delete a file from our primary Drobo B800fs. All we had to do was mount a share from the DR B800fs and drag the file back to the original location on our primary Drobo B800fs with no problem at all. I can't think of how it could have been easier.

In our next test we will power off our primary B800fs and show how to promote the DR Drobo B800fs to be the primary NAS.

Related Reports

Going Wide Area with Drobo Sync
DR for the Rest of Us
Drobo’s Dashing New Dashboard
Upgrading to a Drobo Model B800fs
Another 3 Minute Upgrade
Data Recovery When it Matters Most

Drobo is a client of Storage Switzerland

Previous Entry: “Reliable VMware Storage”

Stronghold Data Chooses Isilon For Scalable, Reliable VMware Storage

Mon, 16 Jan 2012 08:48:54 -0600

Storage Switzerland recently spoke with David Markley, CIO of Stronghold Data, a DR and IT outsourcing company based in the midwest. Stronghold, in business since 2002, has their two main data centers located in underground facilities, (formerly rock quarries) making them impervious to most natural disasters. They feature comprehensive and multiple redundancy of all critical systems, including power, security, networking, compute and storage. Stronghold provides a range of services from simple file-level backup to system-level mirroring and complete infrastructure hosting.

Why Virtualization at Stronghold

While every business must react to their customers’ needs, for some it’s a fundamental part of their value proposition. For Stronghold Data, as an IT service provider, this means being able to make compute and storage infrastructure capacity available to support the DR services that are their primary products. And, they need to make this capacity available to their customers much faster than could be done with the traditional process of buying and deploying servers and storage.

They chose to virtualize their data center environment with VMware because it gave them the ability to decouple the services offered to their customers on the front end from the physical compute and storage infrastructure on the backend. This gave them the flexibility to keep up with their customers’ demands for capacity, safely storing each organization’s data, and making it available quickly.

To make this work, they needed a standard, expandable architecture that could provide capacity required by a service provider. But, it also needed be efficient to support their business model. Service providers need to take advantage of economies of scale in order to keep costs competitive and turn a profit. This means costs must decrease on a per-capacity or per-customer basis as they grow their business.

One of the areas of chief concern is management, especially storage. If not done right, the cost to manage an expanding virtual server environment and its associated storage capacity can ruin the economics of a profitable IT service provider.

Server Virtualization needs Storage Virtualization

Part of the reason to develop a virtual server environment utilizing VMware is to take advantage of its unique ability to move VMs and resources to meet capacity requirements as well as performance service levels. VMware tools such as storage vMotion and Distributed Resource Scheduler help maximize operational efficiency and keep costs under control. But, in order to leverage these capabilities, Stronghold needed to implement a shared storage infrastructure.

Again, Stronghold’s primary business is providing secure backup and DR services. To do this, their software hashes the files they’re backing up and encrypts them for safe transfer to the offsite data vault. This process is fundamental to the way their system handles and protects data, but effectively multiplies the number of files they have to track and store. When they considered storage systems the question of SAN (block) vs. NAS (file) came up. Markley determined the file-oriented architecture of their backup process could be much more efficiently managed with a scalable NAS solution.

Why EMC Isilon

EMC Isilon was chosen for several reasons. Stronghold needed a solution that would scale almost without limit, but more importantly, they needed a system that would scale without disruption. In their business, they can’t move data around to accommodate growth and they can’t go down for a maintenance window in order to add capacity. According to Markley, adding capacity to their former iSCSI-based storage system was “painful”, requiring significant time to allocate LUNs and balance data between existing volumes and new ones.

EMC Isilon enables them to add modules as needed to the back end of their existing storage infrastructure without affecting users on the front end, and without consuming a lot IT’s time in the process. “It took longer to rack the new EMC Isilon modules than it did to add the storage”, said Markley.

Their backup process creates billions of small files, which generates a significant amount of metadata as well. This all must be tracked by the storage system. The ability to index and manage a large dataset with a very large number of file objects is probably as important as raw performance. In this area, Isilon has performed extremely well. The system does the file indexing in the background automatically, allowing normal system use at the same time. It also shortens the actual time required to ingest data, something that’s critical when they add new customers and have to load large data sets into the system.

400+ Days Continuous Uptime

Occasionally, Stronghold needs to replace components, typical for a system of this size. In this area, the Isilon system worked as advertised, with zero impact to users or system performance. The one time a node went down, the system didn’t flinch and the node was brought back up with no user or performance impact. The point is that when components do wear out and subsystems may occasionally go down, highly-available storage systems need to handle it gracefully. In this regard, Isilon has again worked as advertised.

Stronghold also needed a storage system with the name recognition that would enhance their own offering, and make their customers even more confident. Markley said Isilon’s long history and name recognition in the storage space helped in this area. Also, EMC Isilon’s reputation for customer support was a big factor. When they’ve needed support, Markley said Isilon has been excellent, “even the first level support people are great and can handle most of our issues”.

On average, they spend just 10 minutes per week managing the Isilon infrastructure. Isilon’s OneFS file system means no LUN allocation, no data migration and no downtime - truly a set-and-forget solution. “This frees our people up to spend time on internal infrastructure development and customer support.”

Summary

For Stronghold Data, EMC Isilon storage virtualization unlocks the power of server virtualization, which is fundamental to maintaining their business. With VMware, they can view servers as commodities that are allocated as needed and reallocated to support workloads and business levels of their customers - which can and do change without notice. To make that happen, Isilon unhooks them from the limitations of physical storage and provides the characteristics essential to a service provider’s infrastructure: uptime, scalability, efficiency, agility and simplified management.

Related Articles
Server Virtualization & Analytics - One Storage Platform
The Big Data Archive
Is VMware Big Data?
Storage Efficiency Is Key For Big Data
Solve Corporate IT Challenges with Big Data
What is Scale-Out Storage?

Isilon is a client of Storage Switzerland

Previous Entry: “Cutting The Price on 1 Million IOPS”

Dell-Compellent Upgrades To Storage Center 6.0

Wed, 11 Jan 2012 15:44:00 -0600

In addition to announcing their DR4000 disk backup appliance Dell also announced an update to Compellent’s Storage Center software that controls their storage systems. In large part this is an under the covers upgrade as the product moves from a 32-bit operating environment to a 64-bit environment but there are also some important enhancements in terms of supporting virtual server infrastructures.

While Storage Switzerland will get into the details of the importance of moving Storage Center to a 64-bit operating environment in a future article, the key takeaways are; significant improvements in memory capacity as well as substantial performance gains which will come next in the SC8000. The upgrade should also deliver a better total cost of ownership due to capability gains like higher disk densities, better power and cooling and other quality enhancements.

On the outside, Compellent Storage Center 6.0 delivers another progression of VMware integration improvements. Compellent has further taken advantage of the VMware vSphere vStorage API for Array Integration (VAAI). Dell Compellent VMware vSphere™ Storage APIs for Array Integration offers full copy offload and hardware-assisted locking features that extend Compellent’s existing support for block zeroing. These features speed deployment of virtual machines up to 40 percent faster, free up network and hosting resources and can improve storage performance for data volumes shared by multiple virtual machines.

Storage Center 6.0 also includes improved support for Site Recovery Manager version 5 by taking advantage of the storage replication adapter which allows for automated fail-back, as well as new workflows for planned migration and downtime. Compellent continues a support relationship with VMware for SRM allowing all DR hardware and software support to be provided by Compellent’s Co-pilot support center.

Finally Compellent upgraded their vCenter plug-in to integrate with the vSphere 5 Enterprise Manager software suite. This allows Compellent virtualized storage resources to be managed from the vSphere 5 client including the ability to provision, expand, replicate and configure storage. This further simplifies both the server manager’s job as well as the storage manager’s job.

Storage Swiss take

While most of the headlines of Storage Center 6.0 are beneath the covers the capabilities of a completely 64-bit storage operating environment will greatly enhance Compellent’s ability to scale both performance and capacity. This is a critical step and allows them to move beyond a large business storage solution and truly begin making claims for the enterprise storage environment. With 6.0, Compellent should be able to compete toe to toe with the so-called “big boys” of the storage industry.

Recent Briefing Notes
Dell Brings Advanced Backup & SR to the SMB Market
iSCSI Advances to Meet SME Demands
EqualLogic gets NAS and DCB
Demystify Dell’s Dedupe Strategy
Associated Bank Case Study
EqualLogic vs. Compellent

Dell is a client of Storage Switzerland

previous entry: “Cutting the Price on 1 Million IOPS”

Cutting The Price On 1 Million IOPS

Wed, 11 Jan 2012 11:49:17 -0600

Marvell is a leader in the development of storage, communications and consumer silicon solutions, basically they provide vendors with key components so they can deliver complete solutions to end users. Recently Marvell announced the 88N9145 native PCIe SSD controller which promises to slash the cost required to achieve 1 million IOPS on a PCIe card.

This is a module that vendors will install onto a PCIe board and combine it with NAND Flash, a small amount of DRAM for wear leveling tables and a PCIe switch to connect multiple 88N9145 modules. The PCIe switch will then provide a single channel of native PCIe performance to the host CPU.

What makes the controller unique is its modularity. It provides incremental scalability as more of the 9145 controllers are added. This gives a vendor the ability to provide customers an entry-level, mid range and high-end PCIe set of solutions which are based on the same controller technology. This should simplify development costs for the vendor and represent a lower price point for the eventual customer.

For example an entry-level card could use an inexpensive two port PCIe switch and two of the 9145 controllers to provide a small form factor PCIe card for use in a variety of space conscious environments like 1U servers. At the same time a high-end card could be developed with a multiple port switch connecting to many 9145 controllers.

From a performance perspective Marvell claims that the 9145 can generate 93K random read IOPS and 70k random write IOPS. The Marvell driver includes a raid zero like functionality that aggregates the multiple controller modules and their assigned NAND into a single addressable unit. This leads to a scalable architecture that with 16 9145 controllers mounted on a single PCIe SSD board should be able to deliver over 1 million random write IOPS. All of Marvell’s performance claims are based on 4K blocks making these valid numbers for consideration since they are representative of most workloads.

In addition to performance, the controller also supports 128 bit/256 bit AES on the fly encryption, which is something that many vendors have not yet dealt with. Encryption of solid-state devices is going to be an increasingly important issue as SSD popularity rise in the enterprise.

Finally Marvell achieves the scalable performance while maintaining efficiency. Each controller uses less than a WATT of power and are sized such that eight modules can fit on a small form factor PCIe card and 16 on a full-size card.

What does this mean the vendors?

For the SSD vendor this means that they can now be competitive with some of the more vertically integrated solid-state providers and it may mean that some abandon the development effort involved in creating their own controllers via FPGAs. It also means that they can bring the market a wide range of PCIe SSDs that are power and space efficient yet provide performance that meets the demands of almost any data center.

What does this mean for users?

For users this is different than Marvell’s DragonFly products that we discussed in a previous briefing note. The 9145 controller is a component that users will get from other SSD vendors as part of a complete SSD solution. It also means that when those products are delivered the cost to add high-performance storage to their existing architectures has come down in price significantly.

Recent Briefing Notes
Advancing the State of Solid State Caching
What to Look for in Server Based SSD
The Year of the SSD Cache
What is Server Based Solid State Caching?

Marvell is not a client of Storage Switzerland

previous entry: “Advanced Backup and Disaster Recovery to SMB Market”

Dell Brings Advanced Backup and Disaster Recovery To The SMB Market

Wed, 11 Jan 2012 08:48:47 -0600

Dell today announced its first entrance into the disk backup target appliance market in the form of the new Dell DR4000. Targeted directly at the medium-size business, the DR4000 comes with built-in deduplication, compression and replication as part of a unique and all-inclusive licensing program. Dell feels that the medium-sized business market has been underserved by the current crop of purpose built backup appliances (PBBA).

The Dell DR4000 will initially be available in three configurations of either 3.6 TB, 7.2 TB or 12 TB raw capacity. When combined with RAID and the effective deduplication rate customers should expect logical capacities of approximately 35 TB, 77 TB, and 130 TB respectively. This assumes a 15X deduplication and compression ratio which is reasonable for the backup workload.

While users can expect more or less the same deduplication rate with the DR4000 as they would with any of the other market-leading deduplication appliances, what separates the DR4000 from the competition is how well it focuses on the medium-size business. Total cost of ownership is attractive because features such as innovative firmware and an all-inclusive licensing model ensure optimal functionality and the assurance of no hidden costs for desired future features. DR4000 has a simple installation process with full, intuitive remote setup and management capabilities and supports industry leading backup vendors such as CommVault and Symantec.

As mentioned earlier, the replication licenses are also included allowing for WAN efficient 1:1 replication between two DR4000s. This again addresses a key need in the medium-sized business market since many have alternate sites that they can replicate to, but the expense to deploy a replicated backup solution has kept them from leveraging those alternate sites. With the license included on the DR4000, all the customer has to do is buy a second unit and they have built the foundation for a solid disaster recovery plan.

Storage Swiss take

The Dell DR4000 promises to be a contender in the rapidly growing purpose built backup appliance market. More importantly, it is focused on the portion of the market that seems to be underserved. These customers are looking for an affordable solution to migrate from tape or augment their current tape infrastructure. This certainly does not preclude the DR4000 from being used as a remote or branch office solution in larger enterprises.

Almost as important this shows Dell executing on its strategy of integrating the Ocarina Networks technology. It is further evidence that, over time, we will see an end-to-end deduplication strategy that will eliminate redundant data throughout the entire storage infrastructure. Dell is quickly showing both the market as well as its customers and prospects that it can acquire technologies and integrate those technologies in a way that solves real-world problems that the data center is facing.

Recent Briefing Notes
iSCSI Advances to Meet SME Demands
EqualLogic gets NAS and DCB
Demystify Dell’s Dedupe Strategy
Associated Bank Case Study
EqualLogic vs. Compellent

Dell is a client of Storage Switzerland

previous entry: “Server Virtualization and Analytics”

Server Virtualization and Analytics - One Storage Platform

Tue, 10 Jan 2012 09:27:45 -0600

One of the big challenges faced by today’s data center is dealing with the mixed workloads that applications like analytics, back-office operations and server virtualization present to the storage infrastructure. In many instances IT professionals are faced with having to use multiple storage systems to accommodate these different workloads. This of course drives up costs and makes management of the total infrastructure more difficult.

Storage Switzerland recently sat down with a leading online web analytics company to discuss how they chose Isilon to be the single storage platform for all of their disparate needs. This company, like a growing number of others, has all three data types, back office applications, analytics and server virtualization. As we’ve written in the past, Isilon’s unique scale-out architecture allows the use of a single storage volume across all applications for ease of management and increased efficiency, while at the same time maintaining adequate storage performance.

The production IT department for this online analytics company originally went searching for a storage system that could handle its significant CIFS protocol traffic demands caused by storing the results of data collected from monitoring and analyzing Internet-based social media and websites. Data ingest was a significant problem, as was the inability to scale the system from both the performance and capacity perspective to meet the onslaught of data collection and analysis.

They selected Isilon for its scale-out architecture, which allowed them to seamlessly add capacity and linearly scale performance. The system was quickly put into production and the customer demand was satisfied.

Soon after the Isilon system was brought online an initiative in the research and development group was kicked off to examine the potential impact of using server virtualization within that department. The R&D group’s responsibility was to develop and test the software applications that the production analytics department would eventually use. As a result they needed to spin up and spin down servers at will, as well as push the test code beyond the limits of what the normal production environment was likely to do. This meant that R&D often had a more demanding requirement of both server virtualization and its storage infrastructure.

The R&D department manager knew they had to have a storage system that would not add significant complexity to the server virtualization environment. He wanted to make sure that any capital and efficiency gains made because of this move would not be lost in a more complex storage infrastructure.

Based on experience gained at a prior company the R&D department manager was well aware of the challenges caused by block-based storage in virtualized server environments. The difficulty in having to manage multiple LUNS and allocate them to the appropriate hosts was something he wanted to avoid. As a result NFS was the protocol they chose.

As he explored other NFS options, the R&D manager quickly realized that in many cases you have to subdivide those storage systems into multiple volumes in order to maintain adequate performance, which essentially re-created the block level problem described above. Isilon gave him the advantage of using a single volume on a single storage system that could scale infinitely and keep operational costs from growing at the pace of capacity.

Like any startup project however the primary challenge was keeping costs and administrator involvement low. They could not initially go out and purchase a storage system for the task until the project had proven itself. With this in mind the R&D manager turned his attention toward the existing Isilon storage cluster. The R&D team was able to use the free space on this system and cost-effectively add six more nodes to support the server virtualization project. The Isilon cluster proved itself more than capable of handling a raw bandwidth application like analytics, as well as a random I/O application like server virtualization, without impacting either department’s performance.

Implementation

The details of implementation for this particular company are different than others since they already had Isilon in-house by the time the server virtualization project started. In speaking with the owners of the initial project, that implementation went very smoothly and was quickly receiving over a terabyte of analytics information per day.

Since their Isilon clusters had the aggregated performance required to support mixed workloads from a single volume, implementation for the server virtualization team was very straightforward. They were granted access to the Isilon cluster and then simply started storing their virtual images to it. So, at least initially, there was no implementation time, just leveraging a resource that already existed in the data center.

This is a critical differentiator and advantage for Isilon as it demonstrates the system’s ability to support mixed workloads and, at least in the case of this online analytics provider, to support a completely new project with little to no additional storage costs. It also showed the value of Isilon’s single volume concept. Data did not have to be ‘un-allocated’ from current LUNs and re-partitioned to the new project.

Due to the success of their server virtualization project, as well as the ability of the Isilon storage cluster to provide adequate performance and utilize existing assets, the project quickly progressed from the R&D phase to full-scale production. As that happened, nodes were added to the cluster to support the capacity requirements of the growing virtual machine population. In adding this extra capacity, increased aggregate performance ‘came along for the ride’, scaling in-line with the capacity being added.

One of the real unexpected gains that leveraging the Isilon storage cluster created for this online analytics company was how well thin provisioning worked. VMware, their server virtualization platform of choice, does thin provisioning by default when using NFS mount points. Isilon further helps by providing the performance necessary to handle the dynamic allocation of data blocks as the VMs’ volumes grow. This allows the organization to build VMs at size without having to pre-buy the actual storage capacity, which once again saves them money.

The R&D department’s server virtualization project was so successful that the organization’s IT department began to use server virtualization to meet other requirements, such as back office database applications, email and file sharing, further reducing cost and complexity in their IT environment. Based on the success of the analytics department and the R&D department, the back-office IT department also selected Isilon for their primary shared storage infrastructure.

As each of the three environments grew they decided to have separate clusters. However, each of the clusters can be managed and monitored from a single point, once again reducing operational costs as well as keeping capital costs in-line.

Related Articles
The Big Data Archive
Is VMware Big Data?
Storage Efficiency Is Key For Big Data
Solve Corporate IT Challenges with Big Data
What is Scale-Out Storage?
Designing Big Data Storage Infrastructures

Isilon is a client of Storage Switzerland

Previous Entry: “Taming the Unstructured Data Beast”

Taming the Unstructured Data Beast

Thu, 5 Jan 2012 11:05:42 -0600

Likewise Software, headquartered in Bellevue Washington, has expertise in networking protocols, platform interoperability as well as identity and access management. They entered into the storage marketplace to develop a SMB/CIFS stack for several OEM vendors that were looking to support Windows with their storage systems.

As a result Likewise developed Likewise Storage Services that allows these vendors to tie into directory services and provide network protocol support for SMB 1.0, 2.0, 2.1 and 2.2. In addition, they provide NFS 3.0 support with NFS 4.1/pNFS support under development. They can run this platform on top of a supported Linux or Unix based server and are well on their way to developing a REST API so that cloud providers may be able to provide direct support.

While Likewise Storage Services is primarily an OEM product their next product will provide analytics, governance and audit capabilities of the unstructured environment and should be of direct interest to the data center. We all know that unstructured data is growing rapidly. Several reports have indicated that by the end of next year 80% of all data will be unstructured data. Compare this to 2007 when only 50% of all data was unstructured.

While organizations can only do so much in controlling the growth of unstructured data they do need to be able to manage that data. In the unstructured case, management is more than just making sure the right data is at the right place but it is also making sure the data is secure, being accessed by the right person and being shared to the right groups. This is complicated given the heterogeneous environment that is the reality of the NAS environment.

The big concern of data centers right now is to be able to prove that there is some sort of auditing of unstructured data in place so they can satisfy that requirement. This may be as simple as being able to respond to requests like “show me everyone who has accessed this file in the last three months”. Even this sort of detail is difficult to ascertain from most storage management software solutions. Likewise makes gathering this data relatively easy.

The Likewise product will be able to go beyond simple auditing and provide proactive protection of the asset that unstructured data is becoming. For example if an employee were planning to leave the company and decided to copy all their data from their home directory to a locally attached USB drive it would be very good for the organization to know that that’s happening.

Standard storage management software may never catch this event, though the user has access to their files, they are not necessarily accessing something they shouldn’t. It’s just the nature of that access, all at once to multiple files, that should trigger the red flag. Again Likewise will be able to detect such an event providing proactive notification that sensitive corporate data may be on its way out of the building.

Storage Swiss take

With a solid OEM foundation, thanks to their Likewise Storage Services product, Likewise software seems to be on the right path. The ability to track, audit and govern unstructured data is lacking in the data center today. Clearly the sheer mass of data is beyond the ability to have an IT administrator manage manually and most software applications don’t yet provide file level governance capabilities. Most organizations will be interested in using Likewise to meet the audit requirement but then be able to explore the software to take full advantage of what it can do.

Recent Briefing Notes
Optimizing The WAN For Big Data
VMware Specific Storage
Simple DR Solution Eliminates the Questions
Implementing SSD Without Understanding Data
Highly Available Solid State Storage Performance
Integrating PCIe SSD Into the Storage System

Likewise Software is not a client of Storage Switzerland

previous entry: “Optimizing The WAN For Big Data”

Optimizing The WAN For Big Data

Tue, 3 Jan 2012 12:04:12 -0600

Branch office WAN optimization has been a top project for IT professionals the last several years and has created some new IT vendors in the process. New initiatives like large disaster recovery replication, regional VMotion and now big data analytics are driving a demand for data center-to-data center WAN optimization.

Traffic between the central data center and remote or smaller offices is often called “north-south” traffic and typically involves latency between the connections. This latency means lower transfer speeds which allows for the use of less efficient WAN appliances to keep costs down. Data center-to-data center WAN optimization, on the other hand, doesn’t have the latencies common in branch office optimization.

This “east-west” traffic between data centers is often very high-speed ranging from 10 Gb per second to 100 Gb per second. This means the WAN optimization solutions designed for north-south applications may not have the performance necessary for this inter-data center traffic.

Infineta’s Data Mobility Switch is designed for this ‘big traffic’ requirement and can provide 5 to 10 times the bandwidth capacity, IP traffic acceleration and application prioritization, compared to branch office WAN optimization solutions. For inter-data center workflows like production applications, VMotion over distance, large file transfers and now big data, this type of high performance is a necessity. Recently, storage Switzerland discussed with Infineta the impact that big data IT projects are having on the WAN.

Big data, big WAN

Big data initiatives are key focus areas for large organizations today. While many companies are still in investigative or proof of concept phases, there are companies that have moved into production and are seeing significant returns on investment. With these successes come certain problems. As we discussed in a recent article “Designing Big Data Storage Infrastructures” these environments can grow to petabytes in size. But for maximum value the data needs to be disseminated throughout the organization. That dissemination happens via the corporate WAN, which puts tremendous pressure on the network infrastructure.

The first pressure point for the WAN is data aggregation, which involves moving data from a collection point or points to a centralized analytics cluster, often based on Hadoop. Next the cluster aggregates that data and begins processing. Often, this processing is not limited to a single Hadoop cluster or even a single data center. It’s shifted between data centers based on processing capabilities or a desire for the results to be generated closer to where they’re needed. There is also demand on the WAN for DR replication of this big data since, in many cases, the data being collected cannot be re-created. Protecting it and making sure it’s copied is critical.

Finally, the WAN is needed for results distribution. Big data will often generate subsets of data that will be further processed and may even be combined with other structured data warehouses for further comparison. In these cases it’s ideal for that data to be moved closer to where the next round of processing will be done or the results reviewed.

Big data is not just another disruptive technology. It’s a process by which IT can provide information assets to the organization that can be mined for decision support and better profitability. In short, big data, unlike many other IT initiatives, can make organizations money.

The distribution of data is a critical requirement for a successful big data project. That means more is going to be demanded of the WAN. The key though is for big data’s requirements of the storage and WAN infrastructure to be minimized so that the cost of that supporting infrastructure doesn’t outweigh the benefits that big data can deliver.

Being able to provide fully optimized WAN connectivity is going to require products that are focused on that task and have the ability to optimize WAN traffic without impacting the low latency connections between data centers. Companies like Infineta with their Data Mobility Switch are well-positioned to take advantage of this opportunity and allow customers to further exploit the revenue generation potential of big data.

Recent Briefing Notes
VMware Specific Storage
Simple DR Solution Eliminates the Questions
Implementing SSD Without Understanding Data
Highly Available Solid State Storage Performance
Integrating PCIe SSD Into the Storage System
DRFortress Selects Nirvanix to Bring the Cloud to Hawaii

Infineta is not a client of Storage Switzerland

previous entry: “Top Storage Companies To Watch in 2012”

Top Storage Companies To Watch In 2012

Wed, 28 Dec 2011 11:17:19 -0600

2011 was a great year for innovation in the storage industry. We watched new startups launch their products, others refine their messages and mature vendors up their game. There are two reasons to watch these companies. First, they might solve a problem that you are having with storage in your data center and make you better at your job. Second, they may go public or get bought which can be interesting for the industry at large. Here, in no particular order, is our list of storage companies to watch in 2012.

Let me state up front in the spirit of disclosure that many of the below companies are clients of Storage Switzerland and some may be in the future. As always we tried to keep our analysis balanced and neutral. Also missing from this list, intentionally, are the “big boys” (EMC, HP, Oracle, IBM, NetApp). We will discuss them in a future article.

NEW

STEC - A major company to watch in 2011, STEC will increase its importance to the industry in 2012 thanks in large part to its CellCare Technology which lets them improve the life expectancy of Flash NAND solid state drives. CellCare is especially interesting when it is used on MLC Flash NAND bringing near SLC reliability to the less expensive media.

CellCare on MLC is a critical development in making Flash based storage the dominant storage medium in the data center. Flash only storage system vendors have been very creative using technologies like deduplication and compression to blunt the cost burden that memory based storage has. Imagine these systems with STEC CellCare on MLC NAND. The combination would drive Flash only systems down another price band making them more appealing to a broader range of customers.

Scality - Most cloud storage vendors that provide cloud storage systems are focused on providing cost effective storage solutions that are primarily for fixed content use cases providing "ok" performance but not typically production class performance needed by online applications. Scality is different in that it has built in tiring to move data between classes of storage as needed. Scality even provides different data protection schemes based on the activity of the data set. Active data uses replication which maintains better performance and then uses erasure coding for more space efficient protection of less active data.

Scality's model proved itself in 2011 by being the primary storage architecture for several webmail application providers. In 2012 we expect Scality to broaden their architectures appeal to any data center looking for a data storage system for applications with 100TB + of mixed access data.

GridIron - GridIron’s TurboCharger is a SAN-attached read-caching appliance that’s designed to close what the company is calling the “server-storage I/O gap”. This unit, which contains DRAM and up to 6.5TB of MLC flash, produces 100K IOPS and 1.6GB/s of throughput performance. Multiple systems can be clustered to scale performance and provide high availability. Designed for use cases involving large, performance critical block-storage applications, like big data analytics and production OLTP, GridIron’s caching is also block-based.

This means it can’t leverage file system metadata to understand how blocks are associated with each other or how they’re related to a particular application. It requires a much more complex process to determine caching candidates, but also produces a more sophisticated set of characteristics about the behavior of data objects. The result is an intelligent caching algorithm that can actually predict which data blocks will need to be in cache before they’re actually requested by an application.

According to GridIron the TurboCharger can accelerate storage access in critical applications by as much as 1000x. For applications like big data and ‘big virtualization’, which are generating enormous requirements for storage IOPS, GridIron’s approach looks very appealing.

Avere - Avere may be the most rapidly evolving company on our list. What started out as a caching solution has really turned into a performance NAS solution that leverages legacy NAS to provide storage services like snapshots and replication. This allows Avere to focus on optimizing the performance of their system instead of having to re-invent the storage services wheel. It also provides the storage manager the benefit of improving the performance of their storage infrastructure without having to throw out their legacy hardware investment.

As we discussed in our article “Storage Performance Sprawl” data centers are having to use multiple NAS heads, with faster and more expensive disk drives to keep up with the demand of the data center. Avere uses a cluster capable appliance that combines DRAM, Flash and high speed disk to accelerate both reads and writes to and from legacy NAS storage.

In 2011 they added a global name space so that they could also consolidate the legacy NAS appliances behind a single mount point if the customer desired. In 2012 we expect Avere to push the performance envelope even further and potentially enable interesting cloud services. Since IT spending will only be up slightly in 2012, solutions that can extend the life of existing storage services like Avere’s should be hot tickets.

Nasuni - One of the most talked about topics in 2011 was Cloud Storage. Common questions where what is it and more importantly what can I do with it? Nasuni provided one of the earliest and best answers to the what can I do with it question. As you can see from our 18 month test drive of the service they have proven to be a rock solid provider of outsourced file services.

Nasuni leverages a on premise virtual NAS appliance with embedded cloud services to provide a complete file service offering that provides file sharing and integrated off-site data protection. Late in 2011 Nasuni added a global collaboration capability to their solution that leverages the cloud to distribute data to virtual NAS appliances to every office that a company might have.

We believe 2012 will be a significant year for Nasuni since much of the cloud missionary work has been done and general resistance is fading. IT Managers are going to be looking at how to solve problems, solutions like Nasuni the wisely embed the cloud will likely be on the short list.

D-Link is a company most people recognize from the networking industry; but they also have a storage system. It’s one that offers some interesting features and reflects the thought this established company put into their product offering before jumping into a market that’s certainly not short on suppliers. The xStack line of iSCSI storage arrays have the functionality you’d expect for real corporate use cases - hot swappable components, lots of expansion capability (up to 168TB in 14U) and even a volume virtualization package to simplify growth and management.

But they’ve included an ASIC-based processor architecture which produces over 80K IOPS and drives bandwidth up to 1160 MB/s. This processor also integrates networking and storage functions, like TCP/IP processing, leaving plenty of horsepower for capacity expansion. This is something you don’t always hear from array vendors making products for the small to medium enterprise market where ‘commodity hardware’ is the favorite selling point. D-Link also allows users to plug any compatible disk drives into their arrays, something which a company may not decide to do, but it’s nice to have the option.

Bridgehead Software - The growth of data storage is acute for most companies, but it can be even worse in the medical industries where compliance is forcing organizations to keep data and image files (growing with each increase in resolution) for very long periods of time. Bridgehead’s Healthcare Storage Virtualization (HSV) is a storage-agnostic solution that allows a company with healthcare-specific data, like PACS system files, to consolidate their existing silos of storage into a single archiving infrastructure that includes their traditional corporate data sets.

The software virtualizes storage from multiple systems on the back end and enables a company to consolidate their existing infrastructure and replicate it off-site for DR purposes. The HSV technology compresses, de-duplicates, encrypts and archives data from medical and non-medical sources on the appropriate storage platform, including the cloud if desired. The Bridgehead HSV helps bring the economics of consolidation tiered storage and the cloud to many in an industry with some of the largest data sets, the longest retention times and the most regulations.

Virident - PCIe based Solid State Storage is quickly becoming a crowded market and it is important for companies to prove that they are just a bunch of memory on a card. Virident’s differentiation is their ability to provide incredibly consistent performance across a wide variety of workloads and a wide variety of capacity levels.

To deliver this high level of consistent performance requires both efficient software design and powerful hardware design. Virident’s efficient Flash management software architecture was originally designed for more stringent NOR Flash market, instead of the NAND Flash Market. As we discussed in our original briefing report on Virident, their hardware design virtualizes multiple Flash controllers to spread the Flash management load so that when the PCIe board is under a heavy workload condition or it is managing Flash at capacity performance can stay consistent.

Kaminario - Solid state devices (drives, PCIe boards) and Solid State Storage Systems (Flash only storage solutions with storage services features like snapshots and replication) are relatively new comers to the memory based storage market. Appliances like Kaminario and others represent the old guard. These appliances are typically used to solve specific performance problems. Kaminario is unique in that they were one of the first systems to offer complete high availability in their appliance and have a unique scaling architecture that provides a more flexible growth path than other appliance based solutions.

This year Kaminario made another important distinction for itself by introducing an appliance that supported DRAM and Flash NAND in the same appliance where both storage areas where usable. Others had used Flash as a backup for DRAM. While most of these Hybrid systems will likely be mostly Flash, as we discussed in our recent briefing report “Choice is Critical When Selecting SSDs” having access to DRAM for particularly write heavy workloads can be ideal.

Exagrid - The purpose built backup appliance (PBBA) market has certainly come into its own over the past few years. During that time Exagrid has established itself as a leader, especially in the mid-range market where it focuses. Exagrid uses a scale-out architecture to allow mid-range customers to start small and grow their investment as needed instead of having to do a fork-lift upgrade.

In 2012 look for Exagrid to continue maturing its product with tighter integration with software applications like those from Symantec and Veeam. They should also be able to take advantage of processor improvements and disk capacity improvements to be able to scale into the higher end of the PBBA market. Look for 2012 to end with only three serious PBBA vendors instead of the dozen that we have today. I expect Exagrid to be one of those three.

Nutanix - How can a company that advertises itself as the “no-SAN” vendor make our list? As discussed in our two briefing reports on the product, Nutanix is one of two companies we’ve seen (Pivot3 being the other) that are converging the storage and compute infrastructure to deliver a turnkey server virtualization solution that eliminates the need to set up a separate SAN. With disk storage inside each compute node and virtualization software preloaded, the Nutanix environment is providing a scale-out shared storage and infrastructure that is ready to run VMs.

Many organizations that we speak with are heavy on the server virtualization skill set but light on the storage administration front. A solution like Nutanix has to be very appealing to them. and to organizations looking to simplify the storage aspect of a server virtualization project.

Violin Memory Systems - Right in the solid state mix is Violin Memory Systems. They have established themselves in both the server based and memory based storage markets, as well as the shared storage market. They are one of the best at differentiating the solid state appliance or “memory array” as they call it, from legacy storage’s use of SSD. They have been very successful with helping HP combat the intrusion of Oracle Exadata and now have a high availability option that makes them a strong consideration as a stand alone primary storage platform.

Permabit - In its past life as a company that provided a data archiving and compliance product, I’m not sure if Permabit would have made the list of top companies to watch. This is not to say that archiving and compliance don’t continue to grow in importance but more that their off-shoot of that technology, a fully embed-able deduplication API set, could change potentially every aspect of the industry in the next few years. As I am reminded of on practically every new product briefing call we do, the demand for storage capacity is growing. We have reached the point that we can no longer buy enough storage to keep pace and this year’s hard drive shortage is not going to help.

The only viable answer is to be more efficient in how we store that data. There are plenty of techniques, including thin provisioning, cloning, compression and of course, deduplication. Of these, it seems like deduplication is the one that every storage vendor is trying to figure out how to integrate into their storage system. Permabit has the solution, and can speed the path to get the vendor there. We have heard of instances where vendors have integrated the Permabit deduplication API set into their storage software in a matter of weeks.

Texas Memory Systems - This is by far the oldest company on our list. They have been in the ‘new’ solid state storage market since before the advent of flash memory, which is now making it so popular. In the same way that Fusion-io has established itself as the leader in the server based solid state storage market, Texas Memory has established itself in the shared solid state storage market. In this space they have taken their PCIe technology and combined it with a long history of building shareable solid state appliances. The result is a very space-efficient, high performance and now highly available solid state storage appliance - discussed in a recent briefing note.

Nimble Storage - Nimble Storage takes the opposite approach from Pure Storage. Instead of an all flash system you have a minimal amount of flash as a cache to accelerate disk I/O. The difference for Nimble is the use of their custom file system to provide greater efficiencies in how that cache is used as well as to leverage compression algorithms for optimal use of disk capacity. The result is a system that can deliver high performance flash based storage I/O for the bulk of its operations, but also leverage disk for less active data.

Another unique Nimble advantage is because of their capacity efficiencies from compression they can store almost limitless snapshots. Combined with Nimble’s replication capability a Nimble customer may no longer need a primary storage device and a disk backup appliance, instead combining these functions into one. These systems are ideal for traditional medium sized business data center needs that have widely mixed workloads and cannot justify dedicating a storage system for each workload as well as a backup system.

NexGen Storage - NexGen uses a similar mixed media design to Nimble Storage. Their strategy is an automated tiering approach to moving data instead of a caching approach. The result is that the flash tier is used to hold data for days potentially instead of minutes like a cache. NexGen has several unique capabilities. First is a technology choice. They are the first to use PCIe Flash (they use Fusion-io’s product) for the flash tier of a mixed media storage system. Using PCIe Flash provides for more direct access to hot data instead of having to deal with the potential latency of a SAS or SATA backend that’s supporting both flash and HDDs. Essentially NexGen has a high performance “network” for flash based data and a traditional “network” (SAS) for HDD based data.

The second unique capability is their storage software that allows the user to dial in the appropriate networking performance for the given use case. As we discussed in the article “Virtualizing Storage Performance” the NexGen system will allow the user to assign performance in terms of IOPs on a per-volume basis. This can be a tremendous aid to administrators trying to guarantee performance of mission critical applications in the virtual environment that also supports non-critical VMs.

CacheIQ - CacheIQ provides a network based caching solution and is another excellent option for providing a storage I/O performance boost for those companies that don’t have the budget for a storage refresh in 2012. Their solution sits in the network and accelerates NFS traffic from any mixture of storage platforms. This should be increasingly important as NFS adoption becomes more widespread, thanks in particular to VMware’s support of NFS as a storage protocol for hosting virtual server images.

In addition to vendor neutral caching, CacheIQ also provides incredible analytics of how NFS hosted data is performing, what of that data should be cached and what should not. It provides a uniquely visual guide to how efficiently you are using the cache and more importantly how you could be using it better. It then provides the controls that you need to take the information provided by the analytics and fine tune the cache.

SolidFire - As we discussed earlier SolidFire has something we like to see in start ups, a very sharp focus on the market they want to go after. In SolidFire’s case this is providing storage for the cloud compute marketplace. Cloud compute has very large, often random workloads that don’t tend to cache or tier very well, if at all. Flash only solid state would be ideal for these organizations but it has two challenges. First, applying the most expensive storage option (Flash) into the most competitive storage environment (storage supporting cloud compute) can be a non-starter. SolidFire addresses this by providing advanced efficiency techniques to optimize the use of flash based storage including in-line deduplication, thin provisioning and cloning.

The second SolidFire feature is their ability to limit flash performance. We have other vendors on this list that will allow you to ensure a maximum performance level but they are not able to limit the top end performance. While that may sound odd, it’s critical in the cloud compute market. If the first applications and users that come up on a flash storage system have unlimited access to systems performance capabilities, they may assume that the system is “slowing down” when more users come online. In reality these first users essentially got spoiled by unrealistic performance levels they should have never seen. SolidFire addresses this capping performance based on service levels.

Caringo - A provider of an object storage software platform, Caringo is mainstreaming object storage. They bring cloud economics to any organization by providing a complete software based solution that enables large scale, low maintenance storage that remains performance responsive via any combination of x86 hardware.

Compare this to the legacy answer; throw more and more compute power and faster hard disks at the problem. Caringo’s answer is to use a new way of storing and handling those files - as objects. This provides much greater scaling capability and performance on more cost effective hardware that doesn’t degrade as available disk capacity begins to be used.

Marvell - Marvell is a company that many storage managers may not be aware of. They are mostly known as an OEM provider of technology to other manufacturers. An exception is Marvell’s DragonFly product, which may qualify as the ideal add-on to your server for 2012, especially if a storage refresh is out of the budget. The PCIe board as we wrote about in a recent briefing report, provides advanced multi-tier caching technology for servers accelerating all storage I/O, both local and network. Unlike many caching systems it includes a backed up DRAM storage area for caching inbound writes in addition to the normal read caching. This provides a performance boost on the most expensive I/O operation, writes, and makes their handling more efficient.

Nirvanix - This cloud storage provider started life having to compete with Amazon’s cloud offering, which is no small feat. Now they have established themselves as a force to be reckoned with by everyone in the legacy storage market place. Their ability to place the public cloud in your data center solves one of the biggest shortcomings with cloud storage; the latency of the network connection. We have interviewed several Nirvanix customers that have leveraged this capability to deploy applications and capabilities previously unthought of. As organizations begin to fully comprehend the true TCO of their storage assets the Nirvanix model may make sense for an increasing number of companies.

Pure Storage - Pure Storage is one of three solid-state-only storage systems that we are currently tracking in the market right now. As we discussed in a recent briefing report, they, unlike appliances, make the potential trade off in performance to provide complete storage services functionality. Some of those services like thin provisioning, deduplication and cloning are there to specifically offset the price differential. Other features are the luxuries that you would expect in a complete storage system offered by legacy vendors. Flash-only systems are ideal for workloads that simply can’t afford a cache or auto-tiering miss or when the workload is too large to be spread out across multiple tiers. They’re also a good option when purpose built solid state appliances provide more performance than needed and are too expensive in a non-optimized state.

Virsto Software - Virsto is a new breed of storage virtualization company, one that’s dependent on the hypervisor that drives server and desktop virtualization. But instead of replacing the storage services that the hypervisor provides like the traditional storage system does, they are enhancing it. Virsto, for example, uses a log based file system to improve the performance of thin provisioning, snapshots and replication. As we discussed in the video “The Storage Hypervisor” I am very bullish on the idea of the hypervisor controlling more and more of the storage infrastructure.

Fusion-io - They have established themselves as a leader in the PCIe based solid state storage market and created the concept of Storage Class Memory, which we wrote about in a recent article. Fusion-io is positioning flash storage as an inexpensive, high capacity alternative to DRAM instead of an expensive, low capacity alternative to storage. Beyond that they have the technology to backup that strong focus and as a result have become the target of many companies in the PCIe solid state space.

Previous Entry: “SSD Accelerates Growth”

Web-based Email Marketing Provider Turns to SSD to Accelerate Growth

Thu, 22 Dec 2011 09:07:14 -0600

It’s often said that problems an organization faces as the result of success or growth are ‘good problems’ to have. While certainly there is some truth to that statement, for people in IT, growth can still cause tremendous challenges. If an IT department can’t respond to the growth needs of the business, it can end up being a primary inhibitor to the continued success of the company.

This was exactly the challenge faced by ActiveTrail and its IT department. In a recent conversation with Rafael Asor, CTO at ActiveTrail, Storage Switzerland discussed the issues that his team faced as a result of ActiveTrail’s rapid success in the highly competitive, SaaS-based e-mail marketing segment.

ActiveTrail has customers around the world and its customer base is growing rapidly, ActiveTrail focuses on the needs and demands of larger organizations that are using e-mail marketing to increase sales and customer awareness. From an IT perspective, ActiveTrail has the challenge of storing large amounts of customer e-mail address information, sending e-mails and then analyzing the results of that activity to improve delivery and open rates. This combination of large amounts of customer contact information, plus the statistics generated from analytics, requires high-performance storage as well as high-capacity.

The performance challenge faced by ActiveTrail stems from the large amounts of contact data that its customers load into the ActiveTrail system and the need to rapidly analyze its customers’ e-mail campaigns so that statistics can be generated on delivery success, open rates and of course, click-throughs. The issues that ActiveTrail and its customers encountered were slow data access and report processing times in a market where seconds really do matter when it comes to response time.

ActiveTrail’s superior functionality in the e-mail service provider market has enabled them to rapidly grow its customer base. But with each additional customer, the burden on the current storage system to keep pace with the customer demand became greater. The concern was if they didn’t make a change soon, the current storage infrastructure would not be able to maintain a high-level of service in order to keep their current customers happy and meet the new needs of their rapidly expanding customer base.

ActiveTrail’s current system was a high-end, name brand NAS solution which hosted a large SQL Server database of about 1.5 TB, as well as all the analytic data. That system was not keeping up with the customer growth or ActiveTrail’s internal analytic needs. To keep satisfaction high and develop an infrastructure that would keep pace with the on-boarding of new customers, Asor and his team began to research other options to supplement the current NAS system.

The first option was to implement a second independent system, based on the same high-end NAS technology, and then offload the new customer additions to that system. This of course would have a significant start-up cost since they would be duplicating the entire system. And, aside from the purchase cost, there would be additional operational costs since two systems would now have to be managed. For example, balancing the addition of new customer data between the two systems would be an ongoing process.

The other alternative was to consider a higher performing system based on solid-state storage. With this option in mind ActiveTrail explored solid-state solutions from the usual vendors and found that the price of implementing a solid-state system, in many cases, was going to be as much, if not greater, than building a second independent storage infrastructure. It also seemed that most of the architectures would not significantly improve performance. This all changed when ActiveTrail met Kaminario and began evaluating its high performance, yet very cost-effective K2 solid-state SAN storage system.

Kaminario’s K2 is a hybrid SSD SAN system that leverages both DRAM (something that most of the other vendors didn’t even have as an option) and flash-based storage. This combination of DRAM and flash SSD storage allowed for even higher performance because ActiveTrail placed the most active, write heavy, “temp DB” file in DRAM and the rest of the database in flash. Again, most of the vendors didn’t even have DRAM as an option and none of the vendors had the ability to mix DRAM and flash in a single system. As Storage Switzerland pointed out in the article “Choice is Critical When Selecting SSD”, this hybrid model that leverages both memory technologies is ideal for many environments, especially high performance databases.

After careful analysis, ActiveTrail decided to move forward with Kaminario. As has been commented by other Kaminario customers, the most lasting impression they cited was its incredible customer service. The process of installing, testing and, most importantly, measuring the impact of the Kaminario system was very smooth.

Measurement is a critical step that many vendors are ill-equipped to perform. Kaminario, on the other hand, was able to provide a detailed analysis for ActiveTrail. As the chart below shows, in benchmarks against the production storage environment the Kaminario K2 was able to complete database access tasks up to 13 times more quickly.

In addition to the performance gains of the system, ActiveTrail was concerned with availability. They are of course an online, worldwide e-mail service provider and any downtime would be unacceptable. Once again, this is an area where Kaminario is unique in the solid-state storage space compared to other solid-state only hardware providers.

As discussed in a recent briefing report, Kaminario’s SPEAR architecture not only removes single points of failure, but actually provides multiple points of redundancy. Also this redundant architecture is designed specifically for memory-based storage. Many legacy vendors that add solid-state disk drives to existing hard drive-based storage systems have a cost-efficiency and a performance problem when applying traditional data protection methods to next-generation storage devices. The Kaminario K2, since it was built to handle SSD-based performance from DRAM, can also comfortably handle flash SSD storage as well. In addition, the K2 worked seamlessly along-side ActiveTrail’s NAS system and complemented it without any change to the application or database.

This upgrade in performance and reliability will of course improve ActiveTrail’s current customer satisfaction while at the same time allow the on-boarding of an almost unlimited number of customers. In addition, it enhances ActiveTrail’s competitive advantage in the market by enabling deeper analytics and a more feature-rich environment.

Storage Swiss take

ActiveTrail is an ideal example of the next wave of solid-state storage adoption. In this database heavy environment, the performance demands are coming from remote users who connect through high latency Internet connections that are aggregated into a massive I/O storm. Addressing this bottleneck requires a system focused solely on providing performance and is not weighed down with storage convenience features common in mechanical-based storage systems.

These databases will often have a combination of files where some are write heavy and others are read heavy. This mixed workload within a single application can be best served by a solid-state storage system that can provide the right type of memory storage for the right workload.

Finally, unlike the first wave of solid-state adoption, traditional brick-and-mortar database environments, this next wave also requires extremely high availability, something that’s often lacking in many solid-state storage systems. These users are accessing the system 24 hours a day, 7 days a week, so downtime is unacceptable and maintenance windows extremely rare. Kaminario, through customers like ActiveTrail, has proven that it is up to meeting both performance and high availability challenges.

Related Articles
Choice is Critical When Selecting Oracle SSD
Bringing Choice to the Enterprise SSD
The Advantages of DRAM SSD
Will SSD Improve Database Performance?

Kaminario is a client of Storage Switzerland

Previous Entry: “Using Cloud Attached Storage for Backups”