I am tasked with setting up our DR strategy. Currently we are setup for Zone Preference Failover but I am not confident this will be sufficient. I believe that having another independant farm located off site with the same hardware (HA CAG's, HA NetScalers, DB server) would be the best strategy. This way all you need to do is change DNS record to point to DR site and stand up the cold standby license server. I believe there is a 96 hour grace period with ZPF, which in the event of a calamity at my primary site after 96 hours the farm will no longer function, is this correct? Can someone please erify.
I am running about 20 PS4 servers. Also advise if there are other options.
No, with PS4 you have 30 days to run it (due to licensing) but you will not be able to make any changes to the farm. I know there can be issues if the IMA service is restarted on a server after the data store becomes unvailable. Ideally, you could have one farm, use zone preference and failover, leverage sql replication and log shipping, NetScaler HA with GSLB and redundant WI servers. That way you don't have to worry about DNS changes.
Another method that you mentioned is to create a separate farm. The only downside to that is you will have to make changes to the DR farm anytime you make changes to the production farm and have a separate external site. Some companies will purchased two certs and setup two sites. For example - https://primaryctx.mycompany.com and https://secondaryctx.mycompany.com. Doing it this eliminates the need for DNS changes - just need to tell the users.
If you want to talk about this in more detail, send me an email at bcahill@mtm.com.
If you're going to duplicate all that hardware you might as well use those servers in production instead of only on standby. Our DR solution at my previous job had everything duplicated like you mention, but we had everything running full production and we had load evaluators set up to direct users to whichever farm was physically closer to them.
You could use the free version of XenServer to host the entire farm (Store, License svr, WI, PS svrs) and replicate the files regularly from prod to redundant site. If you have Provisioning Server feature (separate or included now with XenApp Platinum) you could create single image of one of your PS servers and then stream them back out to your primary 20 servers as well as replicate the image to dr site where you have a XenServer(s) hosting copies of the primary datacenter. Utilizing either secondary DNS (free but must change user habit) or Netscaler with GSLB as Brian mentioned.
SQL db mirror to protect (summary db, store, edgesight..etc) http://support.citrix.com/article/CTX111311XenServer pool replication http://support.citrix.com/article/CTX117258/
I would create a batch file to backup your datastore to one of your DR Citrix servers, in the event of a DR use this backup copy to make a new datastore and failover the remaining servers at the dr site to that new datastore and u have your farm. then once everything is ok failover them back to the original datastore
take a look at http://support.citrix.com/article/CTX677542 recovery options
Citrix Customer Advisory Community