CUGC User Share Follow-up Q&A: Improving Resiliency of XenDesktop Environment–StoreFront Multi-Site & NetScaler GSLB

By Jake Walsh posted 02-26-2018 08:40 AM

  

CUGC User Share Webinar Follow Up Blog Post: Improving the Resiliency of your XenDesktop environment–StoreFront Multi Site and NetScaler GSLB

First of all, I just wanted to say thank you to everyone who tuned in to the webinar – it was great to see so many people tuned in and the questions were awesome too! In this post I will answer the key questions I couldn’t answer during the webinar. Also, a huge thank you to Dave Brett (@dbretty) for moderating the webinar!

As always – if you have anything else you’d like to know more about, please reach out to me on Twitter @jakewalsh90 – I’d love to hear your comments and feedback. I’ve also tweeted out a survey to decide on the topic for my next webinar. 

Click here for the Presentation Slides. And here for a recording of the session.

 

Question 1: Can a XenDesktop site be stretched if there is a good connection between DataCenters?

Of course – there’s no issue with this at all and it’s quite a common scenario. StoreFront multi-site however, really only comes into its own when you have at least 2 farms. There’s no issue with stretching a Site though and this is common practise I find.

Question 2: Where is the setting for changing Delivery Controller timeout on StoreFront?

This is under the Advanced Settings section within the Store Settings. In StoreFront, click on your Store and then on “Configure Store Settings”. You’ll then see the following under Advanced Settings:

JakePicture1.png 

 

The Communication timeout duration is the value used to determine how long StoreFront will wait for a response from a DDC.

Question 3: Do you have to re-authenticate to StoreFront after a Controller Failure?

Nope – this should just work without any issue. The brokering is done at the time of application launch, whereas the authentication is done before any applications are displayed via the login screen. 

Question 4: How is DNS Delegation handled in a failure scenario?

The main change in a Site Failure scenario is that of the two Name Server records in our Delegated DNS Zone, only the 2nd record (the ADNS IP of our Site B NetScalers) is going to respond. I have tested this with only Site B elements functional (All Site A NetScalers offline) and DNS requests are handled with no issue.

This is in part due to the fact that the lookup is handled by the Windows DNS Server – so a client machine requests the A record for StoreFront-GSLB.jake.lab and this is forwarded by the Windows DNS server to the NetScaler ADNS IPs. See https://support.citrix.com/article/CTX121713 for further details on this.

Question 5: How can we configure resource aggregation for an active/active DataCenter scenario? 

This is configured in a very similar manner to the Active/Passive setup that I demonstrated – but instead of ticking the box within StoreFront for “Controllers publish identical resources” we tick “Load Balance resources across Controllers”. This will spread users out across the Active/Active DC Setup.


#XenDesktop
#UserShare
#StoreFront
#NetScaler
#GSLB

Comments

03-24-2018 12:32 PM

Multi-Site Lab Setup

Hi Jake,

Thank you for your note, I got busy and didn't check mycugc.org for your reply.

I have managed to build out my lab complete with segmentation of the two sites being in different subnets to try and simulate a future state of a production environment. The VMs are built and networked just have to configure the Citrix piece. So this weekend I thought I spend sometime on it.

For my first attempt I would like to do configure and test Storefront multi-site without Netscaler Storefront GSLB. Similar to what you demonstrated and I had a couple of questions for you.

On my first attempt - I would like to go with a "single" XD site - Primary and DR Zone and my VM placement design looks like the below:

Primary Zone (Site A)
2 x SF   (storefront.lab.local)
2 x DDC (ddc01.lab.local, ddc02.lab.local)
1 x Win7 VDA
1 x WS2016 VDA
1 x SQL Server (keeping it simple for lab due to lack of VM host capacity)

DR Zone (Site B)
2 x SF  (storefront.lab.local)
2 x DDC (ddc03.lab.local, ddc04.lab.local)
1 x Win7 VDA
1 x WS2016 VDA

As per your design I would like to use a single storefront URL so am I good to use the same storefront.lab.local web url for both pairs of Storefront Servers?

Cheers,

Mayur :)

 

 

 

03-07-2018 01:34 AM

Response

Hi Mayur

Thanks for reaching out! :) I kept everything in a single site - but this was more for simplicity. If this is going into production I would split everything out into at least two zones. This gives us the ability to fail over at a zone level as well as making use of the StoreFront features I demonstrated etc. 

For your prod environment I would suggest the use of SQL AlwaysOn - Basic Ability Groups should be OK. 

I am not too sure on the upgrade question I'm afraid - something for me to try in my lab sometime :)

Let me know if you have any more questions

Jake 

03-03-2018 04:08 PM

Multi-Site Lab setup

Hi Jake,

Firstly I'd like to thank you for your knowledge share on how to configure multi-site Storefront HA. The live lab demo was very helpful.

I was inspired by your earlier post @Jake and went on to follow @Julien's multi-site lab config details and so far I have one DC site configured on my ESX host. Below is what I have configured so far. ( going to do a diagram later).

Site A

PFSense FW
Domain Controller
SQL Server standard
2 x SF
2 x DDC
1 x Win7 VDA + 1 x 2012R2 VDA
1 x VPX x 1 (not built as yet)

Before I start on the Site B configuration, I wasn't sure how you have configured your multi-site Storefront/DDC. Did you go for 2 zone or the 3 zones design (1 empty root primary and 2 secondary) or 2 independent sites?

I want to build the lab so that I can gain a hands on understanding of how everything behaves during the different kinds of fail-over conditions.

Ultimately, I am looking to design an On-Prem for a Persistent and Non-Persistent Win7 VDI and 2012R2/2016 environment which makes SQL Datastore a single point of failure because there is no LHC. This design is for a replacement of an existing 7.6 LTSR (single site) environment but this phase is for a (24/7/365) operational requirement. Hence the question I have i my mind is whether to opt for 2 or 3 zones design. We have a Primary Site (currently) but going to be adding a DR and a remote Branch Office site.

Correct me if i am wrong, we can't do any upgrades without downtime in either a 2 or 3 zone design unless we design as a Primary and Secondary (POD) sites and other then License server nothing is shared not even vCenter in the case of a vSphere platform. The most critical workload is going to be the Win7 non-persistent VMs and as a result I have opted for local profiles in my design to avoid the limitation of DFS-R profile replication.

I will greatly appreciates your and others for your input.

Cheers,

Mayur :)