Sep
06

The Future of VMware View: Large Scale View Architecture

While I attending VMworld I sat in the session, Demystifying Large Scale Enterprise View Architecture: Illustrated with Lighthouse Case Studies. The main reason I went was actually to finally meet John Dodge in person. After talking through twitter and Linkedin I wanted to put a face to the name. John was one of the presenters along with VMware View Product Manager, Lebin Cheng.

The reason for this blog post is to share some of the forward looking material and somethings I never really knew about or gave much thought too. If you like some of the items that are mentioned here, please use your VMware Pre-Sales Engineer to contact Lebin and show your support so these features see the light of day.

Todays’s VMware View

Two Kinds of Replication

It makes sense that session information is seating in memory using JMS. I’ve always seen port 4001 and knew it was for messaging but never really but it together. The session state information must live in a world of 4ms or less latency, slow logons and disconnects could occur otherwise. This is the main reason why all the View Connections Servers must live in the same datacenter. Trust me replication problems are nightmare to deal with.

Before you read further – disclaimer-, VMware gives absolutely no guarantees when or even if this will be released.

Future Settings\Options\Features

Multi-Datacenter View today doesn’t really exist today. If you want View pods in different locations you will need to use an F5 to manage the connections and they will be managed separately. Fast forward to the future and you get the below pictures.

    * New Global replication across all connection server (Hoping this is not ADAM!)
    * An Administrator can configure global entitlements from any location
    * Data Store selection optimization(helping with provisioning and rebalance operations.
    * A global URL to allow for the View Client to connect to any desktop. I think something like an F5 will still be needed but should be easier to configure moving forward.

Mutlti-network label for View pool was also talked about. This is huge in my mind if the VLAN can be controlled from the pool creation phase. It will stop the need for multiple golden images for each desktop pool when you really are trying to scale past subnet ranges and pool limits.

Features from vSphere 5.1 should also start to be integrated. Space Efficient Disks and VMFS will be able to support cluster sizing from 8 to 32. For more information read Cormac Hogan’s posts.

A better View API for automation was also talked about. The current PowerShell commandlets take a lot of work to make useful reports so I am all for this. It should also be a big boost if you’re going to try an layer vCD on top of View. I know Xangati uses the current VMware View PowerShell implementation to create their User Experience Profiles. I can’t help and feel sorry for third-party management vendors, they really have to work harder to stay ahead of VMware. VMware can create their own competitive advantage but doing something like this. I guess that’s the name of the game though, evolve or die.

Mirage integration was briefly touched on. I can see this happening quickly now that VMware Mirage website is up. I think the major hurdle for View and Mirage will be at the user installed apps and figuring out where Persona kicks in or doesn’t. I am also very curious to see the impact of 5,000 desktops running Mirage in the datacenter. View\VDI can make any small network problem seem really bad once you place it in the datacenter.

Remember if you like something here and need it in your shop, please reach out to your VMware Pre-Sales Engineer.

Thanks to John and Lebin for a great session.

Speak Your Mind

*