Methods of backing up virtual machines have changed a lot in recent years, and it’s only really in the past two years that the process has matured. Now virtual machine users face the choice over whether to opt for backing up virtual machines with traditional backup apps or with specialised virtual serv er-focused backup products from smaller vendors.
In this interview, SearchStorage.co.UK Bureau Chief Antony Adshead speaks with Ian Lock, service director for storage and backup with GlassHouse Technologies (UK), about the suitability of mainstream backup products for backing up virtual machines and the key best practices to use in virtual server data protection.
Read the transcript or listen to the podcast on backing up virtual machines.
Download for later:
- Internet Explorer: Right Click > Save Target As
- Firefox: Right Click > Save Link As
SearchStorage.co.UK: Are traditional backup apps good enough to back up virtual machines?
Lock: Yes, I believe traditional backup applications are good enough to protect virtual machines. Although perhaps slightly slower to embrace the additional features of newer, smaller startup companies’ products, the traditional apps in general now provide a good set of virtual server backup features.
For example, they all now support the VMware vStorage APIs for Data Protection, which replaced the rather clunky ESX Version 3 VMware Consolidated Backup methods. The vStorage APIs use snapshots rather than clones to make quick copies of virtual machines for backup purposes, while allowing file-level restores for full image backups.
The newer, smaller companies’ products focusing exclusively on virtual machine backup may provide a slightly wider set of functionality, but the fact remains that the majority of companies, particularly large enterprises, still have a mix of virtual and physical server environments, which ideally need to be protected by a single enterprisewide backup product.
Large enterprises have typically invested heavily in consolidation of backup products and on standardising onto a single platform to minimise support and people costs. They have therefore helped to drive the demand for better protection of enterprise virtual server environments from the more traditional backup application vendors.
SMEs, however, [that are] not so encumbered with legacy physical server environments may choose to standardise on some of the newer backup products dedicated to virtual server backup, and therefore gain from the flexibility which they provide.
Some of the advantages of traditional backup products therefore include having one, integrated solution for all backups, both physical and virtual.
They also include the ability to archive long-term retention backups to tape easily and seamlessly; this is a feature which some of the newer products do not yet provide. This means that production of a long-retention backup for off-site storage can be a somewhat convoluted two-stage affair.
We’ve already highlighted some of the disadvantages of the traditional backup applications. They are typically slightly slower to introduce new features and are likely to be more costly to purchase. Those enterprise vendors have to support large R&D, qualification, support and marketing departments as well as large direct sales forces.
We should note here that feature support varies depending on the virtual server hypervisor in question. Support for VMware is obviously widespread with Microsoft Hyper-V and Citrix XenServer less well covered.
While traditional backup applications, I believe, are good enough to protect your virtual server environment, traditional backup methods aren’t. Deploying a standard backup agent onto each and every virtual machine is an inefficient and cumbersome way to protect them. There are much better ways to back up virtual machines now, which we’ll cover next.
SearchStorage.co.UK: What are the key best practices to back up virtual servers?
Lock: Firstly, minimise performance bottlenecks and limit contention for server CPU, memory, network and storage resources. If you install a standard backup client on every virtual machine and then schedule them all to back up at the same time, the physical host will very quickly run out of processing and I/O capacity and grind to a halt.
Secondly, reduce the amount of backup data that actually needs to be transferred from the virtual machines to the backup systems by using features such as deduplication and Changed Block Tracking, which appears in the vStorage APIs for Data Protection and in many of the newer backup products.
Next, keep a close track of host and virtual machine performance during backup, as typically multiple virtual machines will share a single underlying LUN or storage volume. Poor scheduling of virtual machine backups will cause excessive load on those underlying storage volumes, as multiple virtual disks seek to access the same volume at the same time.
Finally, consider allowing your backup application to automatically detect and protect new virtual machines on underlying physical hosts. A typical virtual server environment will see virtual machines being rapidly created, deleted and moved between physical hosts. A rigid backup structure will miss these new virtual server creations and movements, [putting] data at risk. It could be that you enable this functionality for the production environment while disabling it for your development or test environment, where only specific important virtual machines or data sets are protected.
This was first published in December 2011