My top five tips for a successful VMware data backup include:
- Investigate using VMware Inc. VMware Consolidated Backup (VCB)
- Investigate host-based (server resident within a virtual machine environment) point-in-time copy and replication tools
- Investigate your storage array vendor's and backup application vendor's VMware-integrated products
- Investigate using NFS to manage VMFS and VMDK images
- Investigate data deduplication technologies
These tips are termed "investigate" because of the complexity of architecting an ideal VMware environment that would include storage, high availability, disaster recovery, and backup and recovery. While VMware and other virtualization products can dramatically improve server efficiency, it can lead to greater complexity in several areas. The areas I've outlined should help guide you and your organization to implement an effective backup and recovery process for VMware environments.
Conceptually, there are three approaches to backing up VMware systems and applications. The first approach is to duplicate a physical architecture, and replicate it in the virtual environment, i.e., installing and managing a backup instance within each virtual server instance. This is typically the first approach taken, because the concept is familiar, and often the tools, techniques and methodologies are very similar. Although, as one might expect, that scenario may not be the most efficient or cost effective.
Another approach to backup up data is to utilize VCB. With VMware's VirtualCenter, administrators can set up and manage VCB data backup environments. VCBs allow the creation of snapshots of virtual server instances so that these may be backed up. VCB is now included as a part of VMware product offerings above the entry-level ESX product.
Using VCB also provides file-level full and incremental backups for Windows environments, but only full image-level restoration for VMware environments supporting other operating systems. Using VCB allows backup jobs to be created for each virtual machine. The VCB then takes a snapshot of the VMFS file, and mounts this file to the backup server on the SAN. In this way, data doesn't need to move through the virtual instance being backed up; rather the ability to share VMFS data is leveraged to perform LAN-free and server-free backups. Third-party backup software is still used in this scenario, with the agent running on the VCB backup server, rather than the application server whose data is being backed up.
The third method is somewhat of a hybrid and involves the use of the VMFS and VMDKs to back up and restore data. A new feature of VMFS is the virtual machine snapshot. This provides a VMware-controlled point-in-time copy of data that can be used just as array-based point-in-time copies are used for backing up data. An added benefit is VMFS snapshots are guaranteed to provide a consistent image of the VM instance, something that array-based snapshots are unable to do without coordination with VMware or VCB.
Using NAS with NFS may not come to mind immediately, but there are a few reasons why this should be considered. VMDKs are in essence files, and VMFS is a type of file system, hence using file-based protocols fits well conceptually. Another reason to consider using NFS tools is that the performance is on par with iSCSI SAN storage, although slightly slower than Fibre Channel SAN storage. Third, and perhaps most importantly, managing VMDK files with file-based protocols makes life much simpler, and integration with array-based data protection such as snapshots and replication is straight forward and works well. Managing VMDKs in this manner also allows the use of thin provisioning and potentially data deduplication, leading to space efficiency.
Thus, there are several approaches and strategies that may be deployed to back up VMware operating systems and applications. Each approach has its benefits and potential shortcomings. In order to speed backups or for block-based applications, look to deploy dedicated resources for backing up and restoring VM instances. Using VMware's VCB can improve backup performance and offload servers. For environments that require flexibility or to reduce the amount of storage protected, look to leverage NAS devices and VMware integration technologies.
An optimal backup and restore architecture will likely employ several suggestions and strategies in order to meet a variety of hosted application service-level requirements effectively.
This was first published in June 2008