One category however still remains to be understood in reference to the virtualization process. This is the backup or recovery process. Very often, the IT departments decide to deploy virtualization without taking into consideration the way data will be backed up. This is critical. While the data virtualization offers an easy way to back up or recover the data, it also demands a properly laid out back up plan in advance which can handle the things.
On a high level, there are two ways by which the data can be recovered or backed up from the host:
1. By The Agent:
The agents are the software which are responsible for a full back-up on the machine. This also includes the I/O processes. In other words, every single activity been performed in the past needs to be backed up with no loss.
Considering what these software solutions are capable of, in the light of virtualization, these can be installed on two locations in a virtual environment:
- Agents On Server: This means the physical servers. In most enterprises, the CPUs are underutilized. This ensures that the agents implemented on server do not hamper CPU cycles, thereby keeping the host performance intact. These agents are very capable software which take into account everything needed for the recovery. All the past I/O processes along with the full data and applications are backed up in entirety. When run on off peak times, the back-up process puts the CPU to an optimum usage.
- Agents On VMs: This method is CPU intense and not much recommended. This is because the agents when reside in the VMs, compete with the host resources and the performance is directly or indirectly affected. Also, considering that most of the VDI solutions work in dynamic load environment, a CPU intense process add to the unwanted risks which may arise, if the back-up process competes with the load balancing resources.
In this back-up solution, there is no software running on the server. There is an entirely different server used to back-up processes by using snapshots of the activities. There are back-up applications today which are virtualization aware. This means that the interaction with the host server is triggered based on certain parameters like the scheduled time, block size or a specific I/O activity.
This offers a better management and higher CPU efficiency without the threats of load burdening on the server. Still, there are applications which cannot be backed up completely using this mechanism. Those require special handling and thus demand a different back-up process.
Which Back-up Process Is Better?
In a VDI environment, the ideal solution for a complete back-up while retaining server performance is Agent + Agentless Back-up solution.
Benefits of having Agent + Agentless Back-up process:
The VMs and files can be backed-up using the agentless solutions while the deduplication, I/O processes and specialized datasets may use the agents on servers which specialize in the tasks. This rules out the threat on VMs and does not compete with the host resources in any manner. The CPU efficiency and server performance stay unaffected.
However, for the recovery process, the VMs need to be mounted on the server which causes the delay.
For this purpose, the agentless back-up solutions are assisted with the light weight agents. The lightweight agents are the software which specialize in a few tasks, and not full back-up. These agents connect with the VM during the back-up, and help in the computing and mapping of databases. Movement of data is not a part of this process, which is done by the agentless solutions.
Read more on classification of virtualization and virtual desktop infrastructure.