Virtualization Backup Round-up

With VMworld around the corner, it is interesting to note the new and old players within the Virtualization Backup space. The virtualization backup space often includes:

• VM Backup
• VM Replication
• Continuous Data Protection (CDP)
• Storage Hardware Replication

Virtualization Backup includes pretty much anything that will maintain your VMs while allowing your data to be placed elsewhere for later retrieval; such a place could even be the cloud. In this article we will avoid Storage Hardware Replication and discuss only backup software.

	Mgmt Int	Hyper-V	vSphere	Xen KVM	CDP	Replication	File Restore	Restore Testing	Dedupe	Send To Tape
VMware						SRM
CA ARCserve				1	2	2
DoubleTake
PhD Virtual
Symantec					3
Veeam					4
Vizioncore					5	6

^{1 As if VM was a Physical Server
2 As if VM was a Physical Server for VMware/Citrix (requires ARCsever Application HA)
3 Using Symantec AppHA
4 Near CDP
5 Requires vConverter and is Near CDP
6 Required vReplicator}
The key to virtualization backup is reducing the amount of time it takes to make a backup, how much disk space is used, and generally how much network is consumed by the backup process. In all these cases, we are talking about deduplication of data for a single VM as well as many VMs. Every one of the listed players have a form of deduplication. Yet, not all deduplication is the same.
Many dedupe data as it is written to disk, but a few such as PhD Virtual perform Source deduplication which implies you transfer much less data across the wire than other traditional tools. If you tie source deduplication with Change Block Tracking within VMware’s vStorage interface, and the ability to determine if the blocks to be backed up are NULL data (such as 0’s or unused) then you transfer even less data across the wire and increase backup performance greatly. Tools such as Veeam and Vizioncore’s vRangerPro have mechanism to make this work.
But if the ultimate goal is to place the data on to removable media for later storage, you may either have to combine a few tools or buy an all in one such as Symantec NetBackup, CA ARCserver, and DoubleTake. All of them can write directly to tape or other removable media.
Backup is important but Restoration is more important. If  backed up data cannot be restored quickly and accurately then there is a serious problem. This is one of the major issues with using tape backup today. Tapes are too delicate. Many have switched to blu-ray media for this reason.
Another important question to ask is how often backups are tested? Only when needed, once a month, etc. Do you spot test or perform full tests? Since restoration testing is often time consuming, consider automation tools that do VM restoration testing for you. Veeam’s SureBackup tool does this by restoring each VM into a protected space and then allows scripts to be run to test the applications within the VMs. Without these scripts you will not know if the VM boots normally with no errors. The boot test is the start of this type of automation. If the VM does not boot, then the backup failed and should be looked into.
Another part of backup is Replication, where the data is not only stored to be dumped to tape but stored elsewhere such as at a hotsite, another local host, the cloud, etc. Replication is a copy of a VM that is ready to boot without the need for restoration. Replication is a major aspect of nearly every business continuity and disaster recovery plan involving virtualization. But now we are seeing backup tools integrate into Continuous Data Protection (CDP) mechanisms. These CDP mechanisms are there to provide business continuity via application high availability that either use agents within the VMs or via integration with the actual hypervisor.
The addition of CDP functionality into backup tools, improves the overall backup tool. While none of the tools are providing full CDP, they are providing near or close to CDP functionality via their HA, replication, and restoration mechanisms. If you add on the ability of some backup tools to work with physical machines as well as VMs, you gain a single tool for your entire environment. Furthermore, you in effect have a way to easily import physical machines into your virtual environment. Tools like CA ARCserver allow you to restore from one hypervisor yet it stores everything as a Microsoft VHD.
Standards like OVF that capture the entire state of the VM are surprisingly missing from the backup tools.  The problem is that when we make a backup we loose state that surrounds the VM like Security Policies. The use of standards like OVF needs to and probably will increase in the future.
Virtualization Backup has come a very long way since even last year. There are more end-to-end products available, but we still need to improve these tools to support standards and multiple hypervisors.