A big challenge in managing VDI servers is the volume of data to host. Each VDI session requires a fully running virtual machine. A virtual machine not only requires the applications that users want to run, but all of the operating system files. An operating system like Windows 10 may require between 15 and 20GB of storage space, and that’s without applications. Add in a common set of applications and each virtual machine image may need 30 or 40GB of storage space.
Deduplication is a technology that saves space by only storing one copy of identical data and using placeholders or pointers for each identical copy. Original versions of deduplication would deduplicate entire files. If you had 10 files that were the same, deduplication would store one copy of the file and then use placeholders to represent the other nine copies. Later versions of deduplication work at the block level and at the chunk level, with a chunk being a collection of identical blocks.
The huge advantage of deduplication in VDI scenarios is that one virtual machine image is nearly identical to the next. Consider the following: An organization has 100 virtual machines to support its VDI deployment, with each VM needing 40GB of storage to contain all operating system and application files. In other words, the organization needs to provision around 4,000GB of storage just to keep the VMs.
If deduplication is enabled on the storage array which hosts the VMs, that 4,000GB figure shrinks dramatically. In theory it might be possible to use deduplication to reduce the required storage to 40GB. In reality, deduplication technologies include redundancy technologies which ensure that storage errors don't ruin the original copy of the deduplicated file. Even with the additional required space for the redundancy software, it’s not uncommon for organizations using deduplication with VDI to achieve storage savings of between 80 and 90 percent.
Reducing storage reduces your costs. This also makes it possible for IT to provision higher performance storage, as not as much of it is required. The end result: faster access, lower costs and more satisfied users.
Underwritten by HPE, NVIDIA and VMware.