Backups in a virtualisation environment take on a whole new meaning, typically complex ( as opposed to the simple outlook that the vm vendors would like to portray ) because now you are dealing with shared SAN storage, vm images instead of files, very specific requirements around backup hardware and setup, 3rd party backup agents and multiple methods for backup from VMware themselves. So much for simplifying your IT infrastructure. But it’s not all doom and gloom – I’ll try and break things down so VMware backup is not as dark as it appears to be.
VCB
VMware provides a Backup/DR API in the form of VMware Consolidated Backup ( VCB ) which allows you to either do backups with the VMware service console tools or integrate with a 3rd party backup tool. Doing backups via the service console is not for the faint of heart so most VMware users will go the 3rd party route. This is done by installing VCB and the 3rd party backup server on a machine ( vm or physical ), configuring VCB to talk to VCenter or ESX, and then setting up your backups via the 3rd party tool’s VMware agent/plugin.
VCB works by generating a snapshot of a vm and then storing it in a specified location on the VCB proxy machine. VCB proxy can run on a physical machine or in a vm. If running in a vm then you need to make sure you have sufficient space to store the vm images that are generated at backup time – take 3 to 4x the space of the largest vm you have as a rule of thumb.
VCB Modes
There are 3 main modes of backup using VCB proxy depending on your hardware setup/design. Let ‘s take a look at each of these in turn and the pros and cons.
SAN mode
ESX needs to have its VMFS stored on shared FC SAN storage or iSCSI shared disk. Backups are offloaded to a physical VCB proxy which is also connected to the shared storage. In this mode, the LUNs exposed to the ESX servers need to be exposed to the VCB proxy machine as well.
SCSI Hot-add mode
In the SCSI Hot?Add mode, you set up one of your virtual machines as a VCB proxy and use it to back up other virtual machines residing on storage visible to the ESX Server that hosts the VCB proxy virtual machine. This mode eliminates the need of having a dedicated physical machine for your VCB proxy and does not require you to expose SAN LUNs to the Windows VCB proxy.
In this mode, you can use Consolidated Backup to protect any virtual disks on any type of storage available to your ESX Server host, including NAS or local storage. The only exception is that it does not back up any disks of the virtual machine that has any independent disk, Physical Compatibility RDM, or IDE (This is applicable to ESX 4 and ESXi 4.)
Consolidated Backup creates a snapshot of the virtual disk to be protected and hot adds the snapshot to the VCB proxy, allowing it to access virtual machine disk data. The VCB proxy reads the data through the I/O stack of the ESX host. If the ESX servers only have local disk then you need a VCB proxy on each ESX host.
LAN ( NBD ) mode
In this mode, Consolidated Backup uses an over-the-network protocol to access the virtual disk. The ESX Server host reads the data from the storage device and sends it across a network channel to the VCB proxy. A limitation is that vm’s should not be over 1TB in size.
You can also use a vm for this mode to host VCB proxy however remember the issues relating to tape connectivity. A separate ‘backup’ network can be used between a physical VCB proxy and the ESX hosts to split normal and backup traffic.
Types of backups
There are also a number of various ways of doing the backup:
- image level where one backups up the vm and all its associated files
- file-level ( via VCB or 3rd party agent ) where you backup the contents of the vm at the file level; this can be combined with full and incremental/differential backups ( note that file-level backups via VCB are only supported on Windows platforms )
In addition you can also load a backup agent in the vm and treat it as a normal backup client. Most likely you will use a combination of image-level VCB backup with agent-based file-level backup. Note that VCB can not be used to backup clustered vms and you will need to use 3rd party cluster aware backup tools for this scenario.
Tape connectivity/VTL
Tape can be connected in a variety of ways for the purposes of VMware backup, mostly depending on the storage mechanism you are using with VMware.
- connect the tape device to a standalone VCB proxy using FC, SAS or SCSI – this gives you the flexibility of breaking backups out of your VMware space
- connect the device to your FC fabric if you make use of shared storage for ESX VMFS
- connect a tape device to an ESX host using pass-through SCSI
In addition, you can use a 2 stage backup solution with disk-2-disk ( d2d ) as the first stage and tape as the 2nd. Using VTL is quite useful as it allows you to send more data flows to the backup server than you have physical tape drives. Tape devices are expensive and as such you may only have one or 2. With VTL, you can simulate as many drives as you’d like and send that amount of flows concurrently to the backup server. You need to be reasonable with this of course especially if you are using Ethernet for your transport.
VMware Data Recovery
DR is a new product with VSphere 4 that does VCB-style backups via the VCenter console and provides for de-duplication of data. There are some fairly severe restrictions with DR, so it’s possibly a tool that you will use in conjunction with VCB and a 3rd party tool.
DR is only a d2d tool so you will need disk in your SAN ( you can also use NAS or iSCSI ) to store the backup data, as well no tape devices are supported. It always de-dupes – you have no option in this regard. DR is an appliance that runs as a vm and provides what is essentially incremental forever backups. You can have up to 8 concurrent backups in flight however you can only use to simultaneous backup destinations. VSS is supported on certain Windows platforms and application-level integrity is provided under certain conditions. Stores can be up to 1TB in size and you can use a maximum of 2 stores with each de-dupe appliance.
Conclusion
Both VCB and DR provide for backups of running vms however they do so under different circumstances. Most often you will use VCB with a 3rd party tool, and perhaps DR when you need fast restore capability for vms. A 3rd party tool gives you the flexibility of backup/restore outside of your immediate VMware environment while the VMware tools give you highly integrated ability within VMware.
