Uncategorized

VMware View Storage Accelerator–The (extended) results are in!

Hi,

as you may be aware, one of the VIEW 5.1 released features is “Storage Accelerator”, this feature is basically nothing short of a game changer because it will significantly reduce the Peak IOPS that up until now you had to size your storage accordingly.

why is it important?

well, the technology is cool but technology only serves the business needs and it the case of VDI, the overall technology is pretty expensive, saving are normally coming from OPEX, not CAPEX (at least not in the 1st year..) so everything we can do to reduce the cost can assist in adopting the technology.

it does it by allowing you to configure up to 2048 MB of RAM per ESXi server that will be used to cache the most used READ blocks, so in operations such as Boot Storm, AV scanning, a lot of the activity will be done directly from the ESXi RAM, the bottom line is that your storage cost just gone down – dramatically!, good for customers and good for us!

here’s how the configuration loos like, one click..

image

attached below are some of the storage saving we have seen while testing this technology, I would like to thanks Jason Ventresco who provided me the data.

The Testing Environment:

2000 desktops in 2 x 1000 desktop pools – Win7x32, Office 2007. For this test the desktops were 1 vCPU and 1 GB ram

View 5.1 RC (all that was available at the time as these tests were run around the end of April) and vSphere 5 Update 1

14 Cisco C260 rack mount servers – 2 x 10 core E7-2870 processors / 144 GB ram / 2 x 10 GB links each (desktop hosts)

Servers and array connected via dedicated Nexus 5548

VNX5500 with:

30 SAS pool – Linked Clones (12 file systems for linked clones, 2 for replica images. Replica images are on NFS mounts FS1 and FS2)

32 NLSAS pool – User data (2 file systems for view persona management, 2 for user home directories)

4 EFD  – FAST Cache

Storage layout/config is pretty much an exact doubling of the layout in this document (1000 desktop RA, only difference is that it used a VNX5300): http://www.emc.com/collateral/software/technical-documentation/h8305-infra-view5.0-vsphere5.0-ra.pdf

Additional SAS are present for infrastructure servers

The Results:

Storage Processor(s) Utilization, this has been the usual suspect to look at when peaking, the results with / without are nothing short but astonishing!

clip_image002

clip_image004

clip_image002[5]

These are the three scenarios where CBRC provides the most benefit:

clip_image002[7]

At the 101 minute mark above all desktops are running a steady state Login VSI workload, which is where the benefit disappears due to the transition to the write heavy workload.

clip_image004[5]

clip_image002[9]

there you have it, I hope you are amazed as much as i am, your total VDI solution pricing just gone down!

Categories: Uncategorized

1 reply »

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s