The D2D Proving Ground


In many data centers, the disk in a disk-to-disk backup strategy takes a lot of abuse. The typical backup environment's mission is to move as much data as fast as possible across the network and onto the backup disk. One of the challenges is that this high performance comes in bursts, meaning the backup disk sees high write traffic for a period of time and then slow write traffic. This high-write, ‘bursty’ type of performance is a difficult technical challenge for any disk storage system and as a result, disk that is used as a backup target is pushed to its limits.


In addition to the performance aspects of the backup process, duty cycles for these drives are usually quite long, with backup jobs running more than six hours in most cases (sometimes all night). The result is that the disk systems are run at variable speeds for an extended period of time.


Failure here, while not catastrophic since it’s the second copy of data, can be very painful. For example, if the backup disk is being leveraged to implement a backup strategy that includes more incremental backups (instead of fulls), failure means a much more complicated restore process.


Mistakenly, many users think of disk backup systems as more vulnerable and accept that. The reality is that they actually need to be just as redundant and reliable as primary storage. After all, if your primary storage has failed and you are dependent on your backup for recovery, the last thing you need is for the device you are recovering from to fail as well.


Systems like those from Nexsan Technologies are proving themselves to be fast and reliable for D2D use and now are moving their way up the storage tiers. In most cases the disk systems used in D2D backup are also more cost-effective than what was used in the other more primary storage tiers in the data center. The potential of a lower cost storage platform that can survive the performance and duty cycle demands of the D2D environment are now being looked at for other uses in the data center. IT professionals are moving disk backup systems upstream.


Moving upstream


Moving disk backup systems upstream is essentially moving it up the data ‘value tier’ as confidence grows in the solution. In the end, many customers wind up using what started as a disk-backup solution as their primary storage. Customers, especially those in the SME space but also larger companies, are now finding that this approach can lead to a single storage platform for most of their storage needs.


For years customers have been told they need specialized storage platforms for specialized needs. This thinking gave birth to the information life cycle management (ILM) storage fad of the early 2000's as users looked for ways to lower aggregate disk storage costs. It may be time to challenge that thinking as a common disk platform can be considered for all storage applications in the organization.


For some customers and some data it’s indeed accurate that specialized, high performance is needed for certain applications. But even within this group of customers, often the bulk of the data can be stored on more general-purpose platforms and a good source for these is the D2D backup proving ground.


Archive


The first step up the value chain often is to use the disk backup system as a disk archive as well. Disk archive provides the ability to move data off of the specialized primary storage systems and onto significantly less-expensive disk. There is often a $7 to $10 per GB price difference between specialized, high-performance primary storage and disk archive. Slowing the rate at which primary storage is expanded just makes sense in any economic situation.


This delta, plus disk archives’ quick access to old data when needed, tends to make customers more aggressive when migrating data from primary storage - as compared with disk archive's predecessor, tape. Tape’s one perceived advantage is that when it’s not actively being used, it is very power-efficient. Systems like those from Nexsan include the ability to power down drives when not in use, gaining much of the same power efficiency advantage as tape without the recall performance loss.


The case for archiving may be strong enough just considering its ability to slow ongoing primary storage costs. It also provides a foundation for an organization to protect itself from data related litigation. Companies like Nexsan can add compliance and retention capabilities to their systems by creating specific polices to retain certain types of data and by providing WORM storage on disk to certify it’s not been altered.


Secondary Storage 


As confidence in the storage system continues to grow, its use can be further expanded to provide secondary storage services. A frequent use of this type of storage is user home directories and group shares. Clearly, reliability is as important at this tier of storage as any other. And while performance is still important, this type of data doesn’t require the performance that can be delivered by specialized storage. Once again, shifting this type of data that is important and frequently accessed but does not need high performance to a more general-purpose platform further shrinks the ongoing requirement of specialized primary storage.


Primary Storage


The final area to explore is that of the high performance storage platform itself. Often the initial focus of primary storage is speed. Should it be? Even if the system could deliver 1 million IOPs, it has no value for a data structure that only demands 100K IOPs. This is not so much a function of how large the organization is, but more the nature of the applications that the organization uses. If the application workload only needs standard performance, why pay for a premium for performance that will never be used?


While the performance capabilities of primary storage often garner much of the attention, the reality is that performance of general-purpose storage is more than adequate for most of the data in many organizations. Most often the primary storage vendor decision is made on the reliability of that system and the quality of the organization that stands behind it.


Other than personal experience or company reputation, there is little that an IT decision maker can do to really verify the reliability of a new storage vendor’s products, unless they have been able to watch the hardware and vendor in action as they move the disk backup system upstream. Each step of the way up the data storage tiers, the systems can be put through their paces and a predictive analysis of reliability and vendor support can be made. By the time the decision is made to move critical data to the general purpose platform, that platform has been well vetted, potentially more so than the primary storage that it is replacing.


This makes the backup–to-disk hardware selection more critical. In such cases, you should really look for a quality solution that could eventually be moved upstream to primary storage.


Why can’t storage be the same?


If the disk backup system is moved all the way upstream, this leads to the eventuality that all or most of the organization’s storage is the same and that there are no physical tiers of storage - almost an ‘anti-ILM’ strategy. To be clear, there are exceptions. But many organizations can store most or all of their data and applications on one type of storage system that is easy to use, affordable and provides enterprise-class reliability.


This practice also best enables a strategy that we discuss in our article entitled "Archiving in Place", which leverages the power efficiency of MAID storage to allow older data to age itself out as new storage is added. Approaching storage this way has tremendous efficiency and value to an organization. It reduces acquisition, integration and management costs. It also dramatically reduces complexity while it broadens the meaning of the term “value beyond just financial” to include efficiency.


What business would not be better positioned to survive and grow in the current economy with a disk storage strategy that delivers real CapEx savings while it simplifies the infrastructure? Thanks to a common storage platform that is robust and flexible enough to serve most or all of their data and applications, from disk backup to primary storage, this disk strategy exists today.