Slashdot Mirror
To Citrix or Not to Citrix?
Saqib Ali asks: "These days, it seems almost any application can be served on a Citrix Farm . However, not all application are best fit for a Citrix environment, and I am sure most IT admins are faced with the tough decision of whether to host an application on Citrix or not. What questions should an IT administrator ask before deciding whether to serve an application over Citrix or just plainly install the application on each desktop? I am NOT looking for the benefits of using Citrix, as I'm very well aware of them. What I want to know is, what criteria should be used in determining whether to use Citrix for an application or not. I just don't want to use technology for the sake of using technology. There should be a methodical way (like a checklist or questionnaire) for determining the feasibility (NOT PROs and CONs) of serving an (any) application on Citrix. Here is a Checklist/Questionnaire that I have come up with. Any more suggestions to add to the checklist?"
The Citrix/Terminal Server platform is usually pretty good, but there are a few things that have caught us out.
:-)
Even in this age of multitasking, some apps still don't share. One timesheet application I know of can't be used on terminal server because it chews up all the processor time it can get it's hands on, which just about kills all the other users.
Even some well known applications can be quite heavy on resources when run on a terminal server, so lesson one is to make sure you know what you'll be running, and how well those apps work in that environment.
The other thing that stung us quite badly when we moved to the terminal server platform is licencing.
One piece of software in particular that we used to rely on suddenly became massively more expensive when we switched from fat client desktops to terminal servers. The licence for the desktop PCs had been a free client licence, but for the same users doing the same thing on a terminal server we were forced to pay for them all to use a server licence for the software.
Needless to say, we dropped that product as soon as we had found an alternative, but it was quite a painful process to do in a hurry.
So lesson two is: work out the licencing implications for everything you run before you start the conversion.
I hope this helps you avoid the mistakes we made.
(Spudley Strikes Again!)
How much access to local system resources does the application need? (Tools for interfacing with system)
How often does the tool need to receive updates?
(More updates means more effort distributing for non-citrix setups)
What is the time taken for each update?
(Simply for manpower running around mnaking changes)
How does the non-upgraded version function with the upgraded system?
(backward compat means less need for single cut migration. Citrix guarantees this)
What are the bandwidth requirements of the software with VS. without citrix?
(It may be that the amount of data you're transfering to/from the server is actaully lighter than the graphical snapshot of the cistrix app.)
My main reason for staying away from Citrix has been the price point. My customer base isn't large enough to justify the expenditure. It makes updates easier, but if the tool has a very easy migration/upgrade path and it isn't an insanely heavy network abuser, i'd go without Citrix.
Bye!
When frustration peaked, they installed a real Windows box besides our HP360, so we ended up with 2 systems; it is cumbersome, but it works.
In my view, all this could be replaced with a single Linux box, OpenOffice added, plus any additional application you might require on top.
> Any more suggestions to add to the checklist?
Lots of good suggestions so far. As a long-time Citrix admin, I would add a couple questions to the fairly good ones you already have.
Will the app vendor still support the app if it is deployed via Citrix? In my experience, this is a really good question to ask up front before deploying the app!
Are your apps mission critical? Do you need high availability for them? Citrix really reduces the cost of deploying and supporting mission critical applications, but at a price, as another poster in this thread rightly pointed out. If you don't have the numbers to get the bulk discount rate, Citrix may not save you all that much.
Finally, one that probably doesn't need to go on a checklist, but one that you should ask yourself anyway. Are you willing to work with Metaframe, with your users, and with the app vendors to make it work? We have a nice stable Citrix environment here at the rocket ranch, but we worked with our vendors, and with Metaframe to make it that way. It didn't happen overnight, and I had to take several Citrix training courses before I got really comfortable with Citrix. It was worth it, but that is just my environment -- ymmv.