Redundancy Opinion
So I am setting up a better Citrix environment here at work. I have 4-6 servers I can implement to create a good sound environment.
My plan was to have 2 back end servers and 2 front end servers. The back end servers are load balanced by Citrix itself, and the front end servers were going to be web servers load balanced with NLB. However, I wanted to put a web server in our backup location for physical redundancy. Come to find out, NLB doesn't support servers in different subnets so I don't have that option.
So what is the best setup in your opinion?
1 back end server in main building with 1 back end server in fail-over location. One front end server in main building and one front end server in fail-over location but turned off and used only for emergencies.
OR
1 back end server in main with 1 back end server in fail-over location. Both front end servers in main with NLB configured.
OR
Another solution.
I don't have access to any hardware load balancers. And having a server turned on only for emergencies makes it tough because I would have to change a DNS setting to point to the IP of that machine... Plus if I wanted to make a change to the production front end server (patches, web site change) then I would have to fire up the backup and put the change on it, then turn the server off. What do you think?
My plan was to have 2 back end servers and 2 front end servers. The back end servers are load balanced by Citrix itself, and the front end servers were going to be web servers load balanced with NLB. However, I wanted to put a web server in our backup location for physical redundancy. Come to find out, NLB doesn't support servers in different subnets so I don't have that option.
So what is the best setup in your opinion?
1 back end server in main building with 1 back end server in fail-over location. One front end server in main building and one front end server in fail-over location but turned off and used only for emergencies.
OR
1 back end server in main with 1 back end server in fail-over location. Both front end servers in main with NLB configured.
OR
Another solution.
I don't have access to any hardware load balancers. And having a server turned on only for emergencies makes it tough because I would have to change a DNS setting to point to the IP of that machine... Plus if I wanted to make a change to the production front end server (patches, web site change) then I would have to fire up the backup and put the change on it, then turn the server off. What do you think?
Comments
-
Mishra Member Posts: 2,468 ■■■■□□□□□□My topics aren't that boring. >_< You all aren't infrastructure geeks in here? I love debating best-practice scenarios.
-
Mrock4 Banned Posts: 2,359 ■■■■■■■■□□I didn't even see your post...so....please bear in mind, I am hardly a novice in infrastructure design. That being said, I would go with the first option. One word- Paranoia. I hate placing all my eggs in one basket, and not having a backup. I'm curious what the true infrastructure experts will say.
Maybe others will jump on now... -
stlsmoore Member Posts: 515 ■■■□□□□□□□Umm out of my league Mishra...give me a year and I'll be able to give you an educated comment lolMy Cisco Blog Adventure: http://shawnmoorecisco.blogspot.com/
Don't Forget to Add me on LinkedIn!
https://www.linkedin.com/in/shawnrmoore -
astorrs Member Posts: 3,139 ■■■■■■□□□□2 x Presentation Servers
2 x Web Interface servers in your main office running NLB
Now assuming that a Presentation Server in your backup office (is this a DR site or a second branch office?) is able to provide service if the main office is offline it wouldn't hurt to add a Web Interface server there too. I would just install it on the PS server though (not a seperate machine or use VMs for all the WI servers). Then configure DNS to pass the WI NLB cluster IP and the backup WI IP as entries for the same alias (just make sure the NLB IP comes up first when you do a nslookup - don't use round robin). -
Sie Member Posts: 1,195Mishra wrote:1 back end server in main building with 1 back end server in fail-over location. One front end server in main building and one front end server in fail-over location but turned off and used only for emergencies.
However I would have the DR Front End Server Online and create the relevant CNAME Record within the DNS Zone as mentioned above.
Main server fails DNS will push traffic to the second server without editing the record and everyone just assumes they are still hitting the alias.
Depends what the load is I guess and how critical it is......... :Foolproof systems don't take into account the ingenuity of fools -
Mishra Member Posts: 2,468 ■■■■□□□□□□Good ideas. I think the best scenario would be to have 2 back ends and 2 front ends with the 3rd web interface fail over server in the DR site. I would definitely push to make these sleeping web interface servers VMs though! I would hate to waste all that hardware.
We also have that "if the building burned down/switch died then we would have bigger problems than just Citrix" thing going on.