In the weeks approaching VMworld 2014, I have been sharing our testing results and deployment suggestions showing the most advantageous impacts of SSDs in virtualized environments. Working on some of the important VDI administrator use cases, I wanted to showcase how VDI administrators can reap the benefits of flash solutions to relieve bottlenecks and obstacles and to highlight use cases where flash can be most impactful. This is my last post in this series (and I look forward to see you next week at VMworld!).
VMware VDI Recompose Experiment Using SanDisk® SSDs
In my previous blog, “VMware VDI Boot Storm Testing Using SanDisk ULLtraDIMM – Impact on VM Availability and End User Experience” and “VMware VDI Pool Creation Testing Using SanDisk ULLtraDIMM – Impact on Desktop Creation” I demonstrated the benefits of including SanDisk’s ULLtraDIMM SSD, a memory channel storage block device, to significantly reduce desktop boot and pool creation time and enable availability of desktops to end users dramatically faster than in an HDD environment.
In this post, I will discuss another pain point, VDI Desktop Recompose, and demonstrate how ULLtraDIMM can help dramatically improve bottlenecks such as in VDI boot storm.
VDI Desktop Pool Recompose
VDI Desktop Recompose is a regular activity an administrator needs to carry out in order to keep all the running desktops up to date with all the latest and greatest patches, and application updates. This is very critical considering the security aspect of a VDI environment. Administrators cannot compromise any delay whenever new updates need to be applied to desktops, especially security patches.
The biggest challenge a VDI administrator faces is the downtime involved. When a Recompose operation (mechanism to update users’ desktop) is on, end user cannot access their desktops. Even if they are doing some work in their desktops, a forced log-off will occur and they will lose their existing desktop session (though this forced log-off can be scheduled so that end users do not lose their working data). End users need to wait until the Recompose operation is completed before being able to access their desktop. Administrators need to ensure that a Recompose completes as quickly as possible so that end user have access to their desktops with the needed updates. Like View Pool Creation, Recompose operation is disk IO heavy and requires the right storage architecture and design in order to avoid longer waiting time.
At a very high level, the below diagram depicts a typical workflow for Desktop Recompose:
What Happens Behind the Scenes
To better understand the technical challenges (and bottlenecks) of VDI Desktop Pool Recompose, let us explore how a desktop pool gets recomposed.
To begin the process, a VDI administrator installs and updates the required patch, software, policies etc. as needed in the already created master image VM and then creates a new snapshot out of it. This updated snapshot is the new master image template containing all the required updates.
Once done, the administrator selects the Recompose operation and points to this new image. Just like pool creation process, a new replica image will be created. Once new replica image is available, the Recompose operation will delink all the users disks (Delta Disk) from the original replica image and point them all to the new replica image. Once successful, it will delete the old replica from the Datastore.
While performing all these steps, a lot of read and write operations occur to storage to make the desktops ready again. The more desktops or number of pools, the larger the size of desktop master image exist, the longer the duration until the desktops updated.
The disk operations for this are more intense and more than what we saw in the testing we discussed in the VDI pool creation blog.
Fig 1: VDI Desktop Pool Recompose
We carried out the experiment slightly differently in order to showcase the impact of Recompose operation. We recomposed 50 desktops for a pool backed by ULLtraDIMM with VMware Datastore and 25 desktops for HDD backed VMware Datastore. The Desktop Pool recompose time is measured from the time the “Finish” button of the pool Recompose wizard is pressed until the time when all desktops are shown as “available” in the VMware Horizon View desktop status in administration window. We used VMware Horizon View VM system time to measure this interval.
We executed the test using HDDs and SanDisk’s ULLtraDIMM SSD to observe the result difference between these two storage platforms.
ULLtraDIMM SSDs connects flash storage to the memory channel via standard DIMM slots, in order to close the gap between storage devices and system memory. ULLtraDIMM SSDs use SanDisk’s proprietary Guardian Technology™ platform to meet the endurance needs of write-intensive and mixed-use application workloads. ULLtraDIMM SSD is also certified as VMware memory channel storage device.
The graphs below shows the impact of each storage hardware to Recompose VDI Desktop Pool VMs
Fig 2: VMware Horizon View Pool Recompose Time (Minutes.)
Fig 3: Total IOPS During Pool Recompose
Fig 4: Write Latency During Pool Recompose
I’d like to highlight two key points about the ULLtraDIMM results:
- We achieved twice (2x) the number of VDI Desktops in half (0.5x) the time to complete Recompose operation
- Write latency is near zero during the Recompose operation
This means downtime for a Recompose operation using ULLtraDIMM is much lower than in HDD environment.
VDI Recompose is an important operation for enterprises deploying VDI solutions. It is a very critical for an administrator to figure out the best approach so that end users will not get impacted by longer downtime.
With VDI, end user expectations have gone high. They want to have their desktops accessed anywhere and anytime. This demands minimal downtime for their desktops. Administrators’ struggles to manage this expectation with the downtime involved during Recompose. With HDD in storage stack, such demands by end users can lead VDI deployment to a failure.
All the studies I shared earlier and in this blog for VDI Administrator operations are tested with very small configuration. The VDI deployments are fairly larger than this, generally ~500+ users, and the impact of these studies will be significantly high as the time gap between HDDs and SSDs will increase as we scale and add more users. For this reason storage administrators are adopting flash for the benefits in use cases such as Boot Storm, VDI Pool Creation and VDI Pool Recompose, and we’re seeing adoption expanding extensively throughout the data center.
As VDI is more widely adopted and becomes mainstream, more application and large size of datasets are being executed in virtual desktops compared to before. This is changing user profiles from being a task worker to a power user in their VDI desktops. As this demands higher storage performance and capacity, flash storage plays a critical role in this phase.
I’m looking forward to discuss this in a future blog post, when I get an opportunity to execute VDI steady state and will showcase how desktop density and user application characteristics are changing with VDI deployment on this blog.
As you head to VMworld next week, take a moment to learn more about SanDisk solutions for virtualization and VDI workloads by visiting our website or read some of my previous blog posts on flash and virtualization that I listed below. If you have any questions, you can reach me at email@example.com, or join the conversation on Twitter with @SanDiskDataCtr
Learn More About Flash and VDI
For reference, the previous posts on VDI administrator user cases can be found here:
- VMware VDI Boot Storm Testing Using SanDisk ULLtraDIMM – Impact on VM Availability and End User Experience
- VMware VDI Pool Creation Testing Using SanDisk ULLtraDIMM™ – Impact on Desktop Creation
Some of our recent white papers and blog posts on other areas can be found here:
- VMware Virtual SAN technical considerations
- VMware Swap to Host Cache Testing Using SanDisk SSDs – Impact on VM Density, TCO and ROI
These blog posts will be followed by detailed white papers to provide helpful guidelines for how to deploy SSDs to achieve greatest performance benefits and cost efficiencies in virtualized environments.