Jump to content
Welcome to our new Citrix community!
  • 0

FSLogix Cloud Cache in a PVS Environment


Stan Svetec

Question

Hi,

Is anyone using FSLogix Cloud Cache in a PVS environment?  I have my profile store on a single SMB share, considering using BVCkup2 to sync to another SMB share and using a DFS namespace to reference the store.  However, I like the 'built in' redundancy that Cloud Cache offers.  Who is using it and what are your thoughts?

Cheers

Link to comment

8 answers to this question

Recommended Posts

  • 0

Cloud Cache will give you actual HA, multiple copies will only protect you from data loss (if you lose a server with active containers mounted, its game over)

 

CC seems to be working pretty well in the new releases, we have customers who are happy with it, and the logon/logoff tax has been reduced with multi threading

 

for PVS environments, move the cache to an alternate drive so as not to punish the PVS cache - outside of that i haven't seen  a lot of it floating around (File Server clusters and VHD locations reign supreme for the most part)

Link to comment
  • 0

I read this about using multiple SMB shares https://james-rankin.com/articles/spreading-users-over-multiple-file-shares-with-fslogix-profile-containers/ which I like the sound of.  My current XA 6.5 farm using UPM has about 9,700 profiles consuming about 1TB of space.  I'm trying to figure out which method will be best suited to a PVS delivered 1912 assuming that all of those 9,700 profiles will result in the same number of profiles.  I've considered CC, but the jury is still out on attaching separate cache to each 1912 VDI.  My current UPM store isn't highly available, so not having a HA FSlogix setup won't be any worse that what I have (or don't have) for UPM now.

Link to comment
  • 0

That's James Rankins post - but I wrote a script to make the approach more robust and safe https://github.com/JamesKindon/Citrix/blob/master/FSLogix/DistributeContainerShares.ps1

 

it's pretty hard to compare HA of containers vs UPM to be honest....UPM is resilient as its file based and a copy streamed (should be) to the local VDA, so if you lose the backend file server, you can survive. Containers however, if you lose that backend server, your entire environment dies in a horrible painful fashion and doesn't come back nicely - it's why we rant so heavily about clustering SMB shares to make sure you are resilient. I never feel OK with leaving a single node file server for Container workloads

 

Cloud Cache offers that nice ground of just having resilience built in at the VDA level, and writing back to two (or more) locations - but it's double the storage cost, and container sizes vs UPM sizes are crazy different

 

J

Link to comment
  • 0

there is some magic you can do with vSphere and clusters and virtual volumes (but I have no idea how it works)

 

you will probably find that if you are all flash based, then you wont notice the cloud cache hit (its less now anyway) and it will just be a capacity discussion instead

Link to comment
  • 0

you would let the script handle it (but you can also set a default starting point if you prefer) - you can run it as many times as you like (sched tasks etc) to adapt on the fly as things change - it will effect new locations for profiles, but existing profiles will be searched for in the existing defined locations first

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...