Multi-Site Scenarios with Citrix and Nutanix

by Shane Kleinert, CTP, S. Florida CUGC Leader

Welcome back! Another Silverton community project has kicked off and we are super excited to get this intro post out to let the community know what we are cooking up! if this is your first time coming across a “Silverton Project,” you may be wondering “what in the heck is Silverton?” Is it a new band? A mountain resort in Colorado? Perhaps a casino? Or is it the most bad-ass community project ever developed?!

If you guessed a mountain resort in Colorado, you’re right – and it’s the inspiration for a series of super awesome community projects. The premise: bring community members together to build Citrix related solutions, and blog it out to the world to share the knowledge and give back to the community! For some more background on the concept, see “High in the Hybrid Clouds – Project Silverton” and “Why Silverton?

 So – what’s this project about?

It all started in the CTP Slack channel.  A few of us were discussing the age-old topic of MCS or PVS (I said OR… it depends!) and the topic of image replication came up in the context of multi-site and DR scenarios. PVS is a relative breeze: copy the VHD between sites, import/assign, and BOOM! – you’re ready to roll. MCS, on the other hand, is not that simple, and there’s not much-published guidance out there on how to make it happen. To get your machine image to the second site, some form of replication is required. This can be done a number of different ways, including storage based replication and software based replication technologies such as VEEAM or Zerto.

Given the lack of information available on how to do this, a group of us decided to do something about it, and this project was born. We’re now in the process of building a multi-site system leveraging the latest releases of Citrix and partner technologies, and we’ll be sharing our experiences with the community along the way.

Why Nutanix?

The Nutanix computing platform inherently provides many benefits when used for MCS-based workloads. A few key constructs built into the distributed storage fabric ensure key challenges around typical MCS deployments are solved. One example would be longer deployment times due to copying Master Images to each datastore during image updates/rollbacks. This challenge is overcome by the use of single distributed datastore. Other benefits include providing data locality (ensuring data is written local to each node), OP Log (for ensuring sequential writes), and shadow clones (bits of the Master Image are copied locally on each node in the background, satisfying data locality) to name a few. Kees Baggerman, fellow CTP, has already provided great write ups on these advantages in the following Posts: citrix-mcs-on-nutanix-ahv-unleashing-powercitrix-mcs-for-ahv-under-the-hood, and  overcoming-challenges-with-citrix-mcs-on-nutanix.

In addition to the wonderful benefits Nutanix brings for MCS, the duo recently released the MCS/PVS provisioning plugin for Acropolis. This enables direct integration with Citrix Studio and Nutanix Prism (their web based management UI) leveraging Nutanix RESTful APIs for seamless MCS/PCS provisioning. For those of you that might not be aware, Acropolis (“AHV”) is Nutanix’s own, very powerful KVM-based hypervisor. Nutanix also supports XenServer, providing two powerful hypervisor options at no additional cost to joint customers.

Thanks to some inside contacts at Citrix/Nutanix, we received early access to the AHV provisioning plugin for Citrix Cloud, so for our testing purposes, we will be using a Nutanix block hosted in Citrix’s Santa Clara data center and leveraging the Citrix Cloud control plane for this system. Pretty slick!

What’s the plan?

Now that you have a background on the project and platform we are building on, let’s discuss the plan! Pretty simple really: we’re going to start by building a multi-site network, complete with limited bandwidth WAN links, firewalls, etc. on NetScaler SD-WAN. Inside of that network, we’re going to stand up a couple of Nutanix infrastructure clusters to provision our workload on, then add in Active Directory and the other supporting infrastructure we need to host desktops and apps. Finally, we’re going to leverage a few of the Citrix Cloud services (including the XenApp/XenDesktop Service and App Layering Service) – plus NetScaler Gateway – to create a bad-ass multi-site system.

Project Objectives:

We’ve got about a million things we can envision doing with this project – there are sooo many interesting angles to explore here! For this first phase, however, we’re planning to keep it to ‘only’ the following objectives:

  • Show how to build a multi-site Production and/or DR environment for seamless failover/failback with Citrix App Layering (MCS and PVS).
    • Including MCS and PCS Scenarios
    • Including Elastic and User Layers
  • Solve for MCS multi-site image replication challenges to enable single instance image management across multiple sites.
  • Share our experiences with NetScaler SD-WAN, and attempt to de-mystify it for infrastructure engineers.
  • Learn and share how Nutanix Acropolis File Services (AFS) can be used to host Citrix Elastic/User Layers as well as PVS vDisks to assist in Multi-Site scenarios.

What’s next?

Lots to talk about, but we want to keep this intro article short and to the point. Once we finish building and testing, we will post blue prints of the configuration, including sample scripts we use to automate image publishing to Nutanix protection domains and update existing MCS catalogs across all sites.

Stay tuned for additional blog posts covering our journey through and test cases that we will develop to support the multi-site scenarios. In fact, we are so excited about this initiative we are going to speak about it at the next XenAppBlog VirtualEXPO on September 22nd 2017! If you’ve not checked it out before, VirtualEXPO is a really sweet online technical conference (hosted by fellow CTP Eirik Haavarstein) consisting of eight JAM packed technical sessions from industry experts from around the globe!

Hope you enjoyed this Intro post and are just as excited as we are about this project! Please feel free to comment below and share any feedback, experiences, ideas, suggestions you might have. It’s always great to hear from other community members. Additional blog posts in this series will be published through the month of September, so stay tuned!

Project Team

I would like to give special thanks to the team members who are helping out with this project: Kees Baggerman, Jarian GibsonRick DehlingerMartin Zugec, and Ryan Revord.

Leave a Reply