Windows 7 Migration |  VDI CAPEX Costs |  VDI Desktop Performance |  VDI Endpoint Security |  Application Virtualization Performance |  VDI Deployment Complexity
VDI Scalability |  VDI User Experience |  Citrix XenDesktop Deployments |  VMware View Deployments |  Struggling with VDI Provisioning
VDI Performance Acceleration

When users transition to virtual desktops, they will demand comparable or better performance than a physical PC. If performance of virtual desktops is poor or degrades as more users are added to the VDI deployment, users will not accept VDI and choose to continue using physical PCs. The root cause of the VDI performance issue is that VDI replaces the dedicated low latency hard drive of a PC with a shared and high latency storage array. As more desktops are added to the VDI deployment, storage contention issues arise from large numbers of VDI desktops competing for a limited pool of IOPS.

Common VDI Deployment Mistake

Many IT organizations that have tried to deploy VDI with a shared storage infrastructure that has a fixed amount of storage IOPS available based on a low numbers of IOPS per desktop. During the pilot, while storage is plentiful on a per desktop basis, desktop performance is excellent. As the VDI deployment scales, more desktops are added without increasing the total number of IOPS and the number of IOPS available per desktop drops. As the number of IOPS per desktop decrease, the response time between the windows operating system and its hard drive located on shared storage degrade as virtual desktops compete for a limited pool of IOPS. The result is poor overall application and desktop performance.

Atlantis ILIO make it possible deliver customers a virtual desktop that is faster than a physical PC, converting users from resisting virtual desktop to demanding virtual desktops. Atlantis ILIO accelerates virtual desktop performance without massive investments in storage infrastructure by increasing the number of IOPS per desktop, offloading IO intensive Windows operations from storage, transparently processing IO locally in RAM and optimizing IO traffic make storage perform more efficiently. The result is a significant performance increase for virtual desktops, which translates into faster virtual desktop boot times, logon, antivirus scanning and overall desktop performance. In addition, Atlantis ILIO also eliminates VDI IO bottlenecks such as boot storms, logon storms and antivirus scanning that can degrade desktop performance.

Increase IOPS per Desktop – Fast Virtual Desktop Hard Drives

With VDI, the virtual desktop response time for common tasks decreases if there are insufficient IOPS for the windows operating system, applications and background services to read and write information to disk. When there are insufficient IOPS per desktop, virtual desktops performance degrades as a physical PC would with a very slow or malfunctioning hard drive. The Atlantis ILIO virtual appliance provides real-time and inline deduplication of IO traffic that results in additional IOPS and boosts the performance of virtual desktops. The impact of Atlantis ILIO on performance is similar to replacing a physical PCs SATA drive with a Solid State Drive (SSD).

Processing IO Locally in RAM

With VDI, the virtual desktop must traverse multiple switches each time it reads and writes data to disk. This process can increase latency to as much as 12ms and decrease desktop performance. Atlantis ILIO presents the hypervisor with a virtual hard drive (.VMDK or .VHD file) that processes IO locally in memory on the same rack (Top-Of-Rack deployment) or the same hypervisor (On Each Server Deployment), making the virtual desktop’s hard drive much faster.

Optimizing IO Traffic—From Random to Sequential

The Microsoft Windows operating system and applications are designed to optimize Disk IO. However, with VDI, the hypervisor converts that optimized and sequential IO that is easy for storage to consume into random IO that decreases storage performance (this is also known as the IO Blender effect). Atlantis ILIO optimizes incoming small block IO into larger block sizes and converts the random IO into sequential IO to accelerate storage performance.