Meeting Virtual Server Data Protection Needs

Meeting Virtual Server Data Protection Needs

Joseph Ortiz, Senior Analyst Today companies are embracing server virtualization because it provides key benefits to a business such as server consolidation, improved resource utilization, improved disaster recovery (DR) capabilities, and lower operational costs among others. However, these benefits of virtualization come at the cost of increased complexity, as workloads now must run in an environment where they are required to share a common pool of physical resources such as CPU, memory, disk, network
I/O, and so forth. Properly protecting data is a major challenge in this new environment. Strategies and tools that worked well in protecting the data in a physical environment now need to be adapted to provide that same protection in a virtual world. Since VMware is one of the most prevalent virtualization platforms currently found in businesses today, this article will focus on data protection strategies for VMware. A virtual machine (VM) is actually a single special file which is designed to simplify virtual machine management and which contains the operating system, applications and data belonging to that VM. As such it provides a high degree of mobility to the VM’s operating environment. This flexibility and mobility across pools of physical resources make the VMs difficult to track and demand special considerations to insure proper data protection of the VMs and their contents. As the number of VMs increases, so does the complexity of the task of properly protecting all that data. There are a number of important factors to consider when devising a virtualized data protection strategy that will meet specific Disaster Recovery (DR) objectives, recovery point objectives (RPOs) and recovery time objectives (RTOs). Some of these are: • Granular file level recovery • VCB support • VM image recovery Application integration
(Exchange, SharePoint, Oracle, SQL- clustered and non-clustered physical and virtual) • Data retention requirements • Centralized management GUI • Ease of VM backup management and tracking • Automatic VM Discovery • Ease and speed of restoring VM data • Backup agent options and methodology • Backup impact on VM performance • Backup with deduplication • Bare Metal Recovery (P2V, V2P, P2P, V2V) • Multiple platform support (Windows, Linux, Solaris) • NDMP support • Reporting capabilities for VMs and physical servers As businesses began to deploy virtualization, many of them used the traditional approach of loading backup agents on the host and guest OSs in order to do file level and application specific backups on the VM and its host. While this method worked, as the number of VMs increased, the backup process imposed an increasing strain on the local host CPU and I/O resources. This had a negative impact on the performance of the applications running in VMs. While VMware comes with VCB, which provides off-host backups of the virtual environment, using associated tools to properly quiesce and restart VM operation, it is not a complete solution. It doesn’t: • Perform specialized application backups • Perform file level backups of non-Windows VMs • Provide granular file level restores to VMs • Provide management, cataloging or archiving of backup files VCB is essentially a framework of scripts, which need to be integrated with third-party solutions to provide these features. There are a number of different backup solutions available today that provide varying levels of support and functionality for the VMware environment through VCB integration and support. There are a few though, which in addition to traditional VCB support also provide additional or extended capabilities and functions beyond VCB, which are optimized specifically for the VMware environment. Syncsort’s BEX 3.1 is a good example of a solution that provides this type of extended functionality for the VMware environment. Among the various protection considerations listed earlier, some major concerns were with VM application performance, backup loads and their impact on I/O and network bandwidth on the VM and its host, recovery granularity, speed of backups and recoveries, and BMR capabilities. Ideally you want a solution that provides in addition to items previously listed: • Fast backups with minimal impact on the VMs and their performance • Fast recoveries of ESX servers and VMs • Fast BMR recoveries (P2V, V2V, V2P and P2P) • Near-instant DR of physical or virtual machines • Non-disruptive cloning of physical or virtual servers to a VM • Snapshot replication • VMotion support • Storage reduction • Automatic tape migration • Flexible backup scheduling • Application verification • Robust job monitoring and reporting on VM and physical backups and resources As with most backup environments today, the speed of backups along with their impact on the client systems is critical. Equally important is the speed with which you can recover a file or an application server or VM. A solution that can deliver exceptionally fast backups and recoveries by using a highly efficient block level incremental backup technology combined with a data reduction process would be a welcomed addition to an organization’s data protection process. For example, Syncsort’s BEX 3.1 can take advantage of block level incremental backup technology to perform a zero impact backup of a VM without impacting CPU or application performance. This is accomplished by creating an initial image snapshot and then from that point forward, only capturing the actual changed blocks and writing them to the backup target. These snapshots can be used for multiple purposes and provide the ability to perform instant creation of a VM without data movement. This is especially important when you are in a DR situation and need to restore a critical application server as fast as possible. For instance, in the event of a disk failure on an application server, this snapshot feature could be further leveraged to allow you to map a virtual machine to a selected snapshot and mount it. At that point the data is live and the application is once again processing data normally. The result is you have your application server back online within a few minutes with no data movement. This same strategy can be applied to bare metal recovery (BMR) situations and allows the recovery of an entire physical server or VM within minutes instead of hours or days. If you extend this example to a total DR situation you can see how quickly you could recover an entire site. Another important feature in a virtualized environment is the ability to un-virtualize a system, moving from virtual to physical machines (V2P) at a later time. Many solutions provide the ability to migrate a physical machine to a virtual one but not all of them allow you to reverse the process. There are several situations in which it could be necessary to migrate a VM back to a physical machine. One example is where a physical machine is migrated to a VM temporarily, while the physical machine is repaired, replaced or upgraded. Once those repairs are complete, you need to migrate the VM back to its original physical host. Another situation is where you’ve migrated physical machines to VMs in a DR situation. Once the problems at the main site are fixed and the site is back up, you need to migrate the VMs back to their original physical hosts. A third circumstance might be where an application was migrated to a VM but later it became apparent that the application was not able to perform well in the virtualized environment. Once again you need the ability to migrate the application back to a physical host. These represent just a few examples of where the ability to migrate from a VM to a physical machine can be very necessary. A comprehensive solution will also provide detailed reporting on the virtual environment and its resources as well as the physical environment being protected. It should show at a glance a variety of data that includes the status of backup jobs, servers, success/failure rates, backup windows, job duration and other statistics that let the backup admin determine whether or not there are any bottlenecks or problems in the backup environment. For the virtual environment it should also show the status of the various VMs, including storage utilization, performance statistics, etc. While the virtualized environment is a complex one with many varying requirements, there are excellent solutions, like Syncsort’s BEX 3.1, available to provide flexible, robust data protection and that is easy to manage.

Wednesday, June 24, 2009