Our Five Tips Backup and Recovery in Virtual Environments

Backup and Recovery in Virtual Environments

Back to the tech stuff if you use VMware and Citrix system.


Introduction
Virtualization is being rapidly adopted, particularly in small to mid-sized businesses (SMBs) where time and money are always at a premium. It brings significant time, money and labor savings in a variety of areas, including procurement, administration, deployment, operation, reliability and recoverability. Virtualization can radically simplify management of the entire environment and enable the SMB administrator to ―do more with less.‖ Moreover, disaster recovery becomes significantly easier once a business has virtualized, provided the administrator adopts newer, more efficient technologies that are designed to work with the virtual infrastructure.
However, like any technology, virtualization brings challenges that can erode its cost benefits and leave the infrastructure less protected than before. In this paper, Quest’s data protection experts offer five tips for effective backup and recovery to help you avoid the challenges that might keep you from fully protecting your virtual assets and infrastructure. You will discover how simple and affordable effective virtual data protection can be, and maximize your investment in your virtualized infrastructure.


Tip #1: Minimize the amount of data you protect
You can reduce the amount of data you back up while ensuring 100 percent recovery by using technologies that filter out unchanged and deleted data.
While tools that utilize VMware CBT (Changed Block Tracking) eliminate the backup of some unnecessary data, CBT does not prevent the backup and restore of deleted data. The Windows operating system uses the unused free space that is allocated, but not used, for data to store deleted files. That deleted data is never removed until it is overwritten to make space for new data. VMs that host applications with frequently changing data can have gigabytes of deleted data. Unfortunately, those files are seen as changed data blocks, and backup tools using only CBT will back up that deleted data. That stretches backup times, lengthens restore times, and overloads your network.
Our tip is to select a tool that does not back up deleted data. That way, you can back up often and with greater granularity. You’ll also save substantially on storage space, backup time, bandwidth and recovery time, enabling you to have better recovery point objectives (RPOs) and shorter recovery time objectives (RTOs).

Tip #2: Maximize backup speed and throughput

Many backup administrators manage data protection for their virtual systems as if they were protecting physical systems; this can seriously reduce the efficiency of virtual asset data protection. For example, administrators often put multiple VMs on a server that would have previously hosted only one physical application. This creates increased contention for network resources—particularly when backups and restores are being performed.
Virtualized systems are different and need different techniques for optimal protection. We recommend you use a tool that allows simultaneous backup and restore to avoid bottlenecks. In addition, use a tool that provides flexible backup methods (proxy, direct-to-target, LAN-free) to fit your environment and minimize workload impact.
To further increase network and system efficiency, choose a tool that eliminates the need for a backup server by sending backup images directly to target storage. This approach reduces network load by eliminating intermediate steps.
Tip #3: Keep your recovery options flexible
 
While agent-based systems have their benefits, they aren’t always most efficient or cost effective for small organizations. When you back up virtual systems with agent-based systems, you typically have to pre-stage your VMs to restore an entire VM. This means you have to spawn a new VM via clone or template, size the memory and disks correctly, name it correctly, and create the appropriate number of virtual disks. Once this is up and running, you must then install an agent, connect to the target, and restore the VM. One alternative to an agent-based system is bare-metal restore routines. However, these are challenging to implement at best, and you may have to maintain duplicate hardware with this option as well.
Fortunately, virtualization brings many simpler and more powerful recovery options. Use a tool that allows you to simply click on a VM to restore it, with no need for pre-staging. Find one that allows you to easily restore files at the file level and to restore application objects. Set up your disaster recovery scheme so you can fail over to a VM on a remote server (either on campus or offsite) with a single click of a button, and ensure the replication is automatically reversed so that once the source site comes back up, you can simply synchronize the changes and failback to source.
What about physical boxes? Almost every virtual environment has some servers that just can’t be virtualized yet. Consider companion tools that work with your virtual data protection tool to offer continuous protection for physical servers. Using continuous protection, you can image physical systems into VMs, which can be then restored to a VMor a physical server. This approach gives you the flexibility to get your systems restored and your business back on line fast.
What about long term tape-based retention? Most organizations already have investments in agent-based software and tape systems. All you need is a single agent with visibility to an archive repository to sweep the archives off to tape. Consider a tool that offers sweep-to-tape integration that can be used with a traditional backup tool. Then if you ever need to recover an old archive, you can simply restore it to the repository, import the manifest, and start restoring files or VMs as you please.

Tip #4: Minimize performance drains
As mentioned earlier, many backup administrators manage data protection for virtual machines as if they were managing separate individual physical systems. Another example of this is deploying backup agents on each VM and running backup jobs in defined backup windows in order to avoid hurting the performance of business operations on the system. Often backups are run during off-peak hours, usually at night.
Unfortunately, this approach has a significant impact on the virtual machine host and VMs. The host system must take on the extra processing load and absorb latency increases due to I/O contention during the entire backup window, slowing all VMs on the host until all scheduled backups are complete. Adding to this impact is increased network traffic and latency due to the increased volume of data traveling to the backup server.
Our tip is to use dynamic resource management to free unneeded resources; when resources are taken only when needed, limited or scarce resources can be shared among processes. You can also reduce performance impact by simplifying your backup infrastructure with a flexible tool that can adapt to your network layout (LAN, WAN, or storage network), shifting the load of data protection operations away from the networks critical to business performance. For even greater benefits, choose a tool that provides flexible backup methods (proxy, direct-to-target, LAN-free).
Reducing the impact of backups on your network, servers and applications will enable you to save on hardware and infrastructure costs. It will also help your current infrastructure perform better so you have room for growth without spending more money. In other words, with the right tools, you can do even more with less.
Tip #5: Protect to fit your needs and SLAs
You have different SLAs and infrastructure for different applications and data. Your data protection solution needs to adapt to fit your needs—not the other way around. Your data protection tool shouldn’t force you to conduct your data protection operations in a way that interferes with your production systems and networks. You should back up only as often as you need to meet your SLAs, in order to minimize effort and load on your production systems and networks.
Therefore, choose a flexible tool that offers a choice of networks and a method to be used for data protection: LAN, WAN, server-less. We recommend an image-based data protection tool because images are very portable, allowing you to recover when, where and how you need to for the greatest efficiency. We also advise choosing a tool with flexible licensing to provide the best fit for your environment while costing as little as possible.
Most of all, choose an architecture that fits the SLAs for your organization. The correct architecture for your business depends on the hardware and setup you have today; there is no one-size-fits-all. Understanding the options here is arguably the most important part of the equation when designing a virtualized disaster recovery system. Regardless of which image-based tool you are using, you need to configure it correctly, which includes, among other things, choosing the correct source method and understanding data flow and proper positioning of targets.
Finally, choose a tool that offers a variety of architectural options for deployment: network-based, direct-to-target, iSCSI, fiber and both ESX and ESXi backups. This will ensure you can set up your backup regime in a way that makes sense for your environment.