As small and midsize businesses (SMBs) virtualize their servers at an increasing pace, many fail to consider the impact this change has on how they do backups – or that it impacts their backups at all. However since many IT administrators who are responsible for backups in these environments would freely admit to not being backup gurus, here are some tips on what features to look for in backup software in order to properly protect and recover your newly virtualized environment.
In the previous entry in this series we examined five ways in which virtualization complicates backup:
- Fewer resources are available to virtual hosts
- VMs use virtual disks
- Difficult to detect new VMs
- VMs can move around
- Granular restores are problematic
These factors give us some indication as to what features organizations should look for when selecting backup solutions to protect their virtualized environments. DCIG has identified five primary features to look for when selecting a backup solution that is designed to protect virtualized environments.
1. VMware integration. Hypervisor integration is the defining factor when selecting a backup solution. Without hypervisor integration the backup software cannot safely and accurately back up virtual machines under the hypervisor’s control. Integration also minimizes, or even eliminates, the need for a guest OS based agent within the VMs under management.
There are number of reasons for this. First, properly integrated backup software will be able to look inside VMDK files to see the data within the virtual disk. This allows the software to improve deduplication and do granular restores of the files inside the VMDK without needing to restore the entire VMDK. This means that users can meet much shorter Recovery Time Objectives (RTOs).
Second, the backup software will have the ability to take snapshots of the VM. Snapshots minimize overhead on the physical host and are less intrusive than guest OS based agents. In addition, it assures that the VM’s metadata and VMDKs are all in a consistent, safe state.
In VMware these functions are accomplished using VADP (vStorage APIs for Data Protection). VADP allows backup software to leverage the VMware tools within their VMs to support both granular file-level backups and shadow copies. When possible, VADP uses guest OS supported tools such as Microsoft Volume Shadow Copies to improve integration When implemented properly, VMs should require zero downtime for backups.
2. VMware vSphere integration. Integration with VMware’s management layer, vSphere, is also critical. Absent the management layer integration, it is difficult to detect and manage the life cycle of a VM. Using this integration, the backup software is informed when a VM is created, modified or removed so new VMs can be automatically included in existing backup schedules and inactive or deleted VMs don’t hold up backups.
The software also is notified if a VM moves between physical hosts (i.e. migrates). This is critical to enforce continuity of the backups as it may eliminate the need to do a full backup when the VM or VMDK moves on a physical host or even moves to a new physical host.
Integration also provides a consistent management environment for administrators. vSphere supports snap-ins to add functionality to the vSphere Client. With adequate integration, backups and restores can be completely managed within vSphere.
3. Application integration. Organizations investigating virtualization will likely focus part of their effort on applications that themselves require integration to perform consistent incremental backups.
Two of the most widely used examples of such applications are Microsoft Exchange and SQL Server. Both applications use large data files that are normally opaque to outside applications, complicating incremental backups. A properly integrated backup software solution will include integration with the applications an organization is using.
This should include the ability to granularly recover tables and/or email messages. If the backup application is integrated with the hypervisor as described in 1 above, then integration with the Volume Shadow Software Copy (vss writer) from Microsoft enables consistent backups for Exchange and SQL Server without an additional agent in the guest OS.
4. Physical host recovery. An often neglected scenario in virtualized environments is the recovery of an entire physical server. Just as in the recovery of a virtual host, the physical host should be returned to a full, consistent state during a disaster recovery scenario.
In virtualized environments this becomes a two-step process. First the physical host is recovered including any hypervisor and management software. This is followed by the recovery of each VM on the physical host. This means that backup solutions must support both physical server backup and restore and integration with the hypervisor for backing up and restoring the virtual environment through integration with both the hypervisor and management layer.
Furthermore, there will almost always continue to be servers in the IT environment that are not virtualized. The backup solution should be able to provide data protection to both those physical servers and the servers that are virtualized.
5. Physical and virtual host replication. One of the strengths of virtualized environments is the ability to quickly create new virtual machines. With the addition of backup software, your environment will gain the ability to replicate point-in-time snapshots of virtual machines. Many administrators will associate this ability with disaster recovery. Other example use cases include creating offsite failover environments, testing and speeding deployments of additional hosts.
Server virtualization fundamentally changes how backup needs to be done. Understanding what those challenges are and implementing a complete backup solution that addresses them is key to ensuring that all of your data is protected, whether it is on virtual servers, physical servers, or in applications.
In Part I of this series, I discussed the five major ways that server virtualization negatively impacts backup.
In Part III in this series we look at how to achieve success in doing backup of VMs without first needing to become a “real” backup expert in order to do so.