Citrix XPE

[IntrepidUsa]

We are curently running Citrix XP with service pack 2 and feature release 2, we are being told that we should switch
to Citrix XPE. I was wondering if anyone is curently running it, or has heard anything about it.

Thank You,
Bryan Morrison
Network/Software Support

 

[ahalecitrix]

My first question would be, why are they recommending that you switch to XPe?

How large is your MetaFrame farm and what version of XP are you currently running?

You can read about the features of each flavor of MetaFrame(XPs, XPa, XPe) here:

http://12.8.192.113/products/metaframexp.asp

The same info is on the above link to Citrix's site, but here is the quick breakdown:

Citrix XPs features:

Product Highlights

Citrix Management Console
Delegated administration
User policies
Citrix Secure Gateway
SSL/TLS encryption
Centralized printer management
Universal print driver
Centralized license management
Citrix Program Neighborhood®
Directory support
Advanced session shadowing
Client time zone support
User collaboration
Citrix NFuse 1.7


Citrix XPa features:

Product Highlights

All features found in MetaFrame XPs, plus...
Advanced load management


Citrix XPe features:

Product Highlights

All features found in MetaFrame XPa, plus...
System monitoring and analysis
Application packaging and delivery
Network management
MetaFrame XP Management Pack for MOM
Enterprise Services for NFuse

 

[JohnTcolo]

Sorry, I posted my reply to the wrong thread, additional info to the first reply:

I think you are getting a lot of bs. We use XPa and the only other thing we might want is the resource manager. And we are getting that from Lakeside software. The resource manager in XPe is merely an earlier version of the Lakeside product.

Here is a link to the Citrix site which compares the various Flavors of Citrix XP, just click on the compare button and you will see the capabiities of each:

http://www.citrix.com/site/PS/products/product.asp?familyID=19&productID=186#top

 

[IntrepidUsa]

We have been told to swith to XPE for performance issues. We run a heathcare program named McKesson, and we are having some serious performance issues. They say we can have 60+ users on any one of our servers and still be screaming along, we hit 44 and we crash! We currently have 6 in the farm and are looking to add another 3-5 servers by July. This is why we want to find out what is going with our performance before we dive into XPE.

 

[JohnTcolo]

You are correct in not wanting to upgrade until the performance issue is resolved. I have about 50 users and only 2 servers. We have made many adjustments to our ICA settings and make extensive use of Load Management and the management rules.

Are you running a lot of WAN clients. Do you have any dialups? What happens to your CPU at the point of the crash? Is it the server or the client that actually crashes?

 

[IntrepidUsa]

We have remote sites from Portland to Rhode Island to Grand Forks to Pompano Beach, so we're all over the map. All of our sites are connected to us through a WAN, but they all use the Citrix Client to connect to the Citrix boxes at corporate. We don't have any dial-ups, not unless their router goes down, then we have them dial-up. When I use the word crash I mean a huge slow down, and it happens on the end user's computer. We've been taking servers out of the farm since last week, 2 at a time. Trying to see if it's a server or trying to determine were are bottle neck is, plus are servers are running Quad Zeons, with 4 Gigs of RAM.

 

[JohnTcolo]

Make sure all of your remote clients are using WAN rather than LAN for the ICA client. We have also modified the farm properties to restrict the amount of memory for both the printers and client memory. Also, there is a tool from Lakeside Software that you can get for a 30-day trial. This tool will allow you really analyze your system and applications. For example, we were able to determine exactly how much CPU our applications used. We used this info to Load Manage.

For example, I now have all Outlook, Excel and Internet Explorer users redirected to only 1 server, the slower one. This has freed up our faster server.

When the clients slow down, what process is taking up the most CPU as shown in the Task Manager on the Citrix box?

 

[IntrepidUsa]

It shows that are SQL servers are slowing us down. Should we look into that Lakeside Company? When we set up all of our Citrix Clints we set them to LAN, we we're told by McKesson that that would be fine, should we be setting them for WAN? You could be on to something.

 

[ahalecitrix]

Can you do some general Win2k performance monitoring? Do you see processor spikes from any particular process.

It's hard to say what is causing the server crash. It could be a variety of different things. I don't think upgrading to XPe just to get the Resource Management piece is the immediate solution. You may decide you would like to have that down the road, but if integrator should be able to diagnose the cause of the problem without using that tool. The management utilities with XPe are great, but intended to be implemented as just that for "management" not for diagnosing a problem although it may be helpful.

I'd check a few things. My first instict tells me to monitor the CPU and memory utilitization of course to see if you can pinpoint the process or processes that spike the CPU time or memory usage if that is occuring. Then try to trace that back to the user and find out what they were doing when the crash happens if possible.

Also check disk space. Disk space can take a MetaFrame server down like nothing else. Here's a few tips.

Make sure the profiles aren't using too much space. These are stored on the System partition so if that partition wasn't made large enough (and in my opinion the 4GB's that Dell or Compaq normally ships their servers with isn't really sufficient for a MetaFrame box. I like to use a 6GB system partition usually for my server builds. Sometimes I even go larger if the disks are large enough.) Remember that if roaming profiles are cohfigured and working properly the user profile will be deleted from the local server when the user logs off. So the disk space utilization may not be apparent after a crash, or if the server slows to a crawl and users start jumping off.

Also, if you have 44+ users on a server and they are doing a lot of printing, that can kill a server by eating up disk space. I'd check that as well. Even if it isn't eating up all the disk space the I/O alone on the system partition can bring a MetaFrame box to a crawl. I make it a practice to move the Spool folder to a separate partition to avoid this.

Just a few suggestions for some stuff to check out. Hope something in there might help diagnose the issue without upgrading XPe.

 

[IntrepidUsa]

I'll check with the rest of the clan to see if we've done what you mentioned. Thank you

 

[JohnTcolo]

Above post has several valid points, follow up on them. And, you should definitely be running as WAN rather than LAN. One of my remote sites was really dragging and I finally got down there and that was the main cause. I asked why in the world they were set that way, and was told someone told them not to use WAN. Believe me, it makes a big difference! Citrix Article CTX283777 covers the differences in detail.

You definitely want to check out the Lakeside Product, it helped us iron things out here. The person at this address can help you out there. carrie@lakesidesoftware.com

Also at the Citrix site, they have a user guide "The Definitive Guide to Citrix Metaframe XP. You should print at least several of the chapters.

Good luck, hope this helps some.

 

[CitrixEngineer]

It is definitely not true to say that RM is a earlier version of Lakeside's product. Although both are based on the same early version, Lakeside have specialised in that product while Citrix have made it a product-specific feature of MetaFrame. The Lakeside product is missing a huge raft of counters that Citrix have put in specifically to monitor MetaFrame servers. On the other side of the coin, the Lakeside product has more features.

Just to clear up that point

I agree that it's not worth upgrading to XPe to attempt to fix a performance issue - the trick is to locate the cause of the issue over RDP. If you can get everything working nicely over RDP, it'll work better over ICA.

I don't know who told you you could get a specific number of users for the Healthcare product - I'd bet that they didn't test every scenario. It is totally unrealistic to predict in advance what any organisation can expect in terms of user numbers from an application running over Terminal Services or MetaFrame.

Quad processor servers do not give twice the performance of dual servers - I normally recommend large numbers of duals over smaller numbers of quads for this reason.

I hope this is helpful

CitrixEngineer@yahoo.co.uk

 

[IntrepidUsa]

A bit of history: We we're running 4 different healthcare payroll systems, and we keep getting more. We decided that we wanted the whole company, Corparate and our branch offices, to run the same program. So we looked around for a Big enough and Powerful enough program to run, and McKesson was it. We've bought servers, tape drives, licences and everything else under the sun to get this program to work. They told us that quad 700 zeons would have no problem running our data base and the application, now they're telling us we need new servers, so we're planning on buying 2 quad 2.0gig zeons, one for the active directory, and one to run the data base. We've been on the phone or they've been up here for weeks now, trying to figure out why are system is slow, they say they run slower servers with more people and don't get a slow down. So we've taking servers out of the farm, changed our our maintenance policy, which helped greatly, and are still working with them to find out were are bottle neck is. We have are next rollout in June, 10 offices, which will push everything we have to the max, then we have rollout #2 about another 10-15 offices in October. So we have to have this figured out and running like a champ before we put anymore people on the system. If we have 200 users now using the system and it's slow, what the hell will happen when we double that!!? If we don't have it running well by June, lets just say we're in trouble.

 

[JohnTcolo]

Correction for Citrixeng, that is an earlier version of the Lakeside product. Just because the source code has been modified does not change the fact that it is still essentially version 1. You need to get your facts in order.

Intrepid, I hear you downloaded the trial version from lakeside, use it and it will help you isolate the problems. And forget about those high-dollar quads, the hardware you have now is plenty-you have software issues. Did you look into the WAN/LAN settings like I suggested.

Take care.

 

[CitrixEngineer]

JohnTcolo - to this discussion, the version number is trivial. The base function - system monitoring and analysis - remains the same.

However, I have discovered recently that perfmon can be used to access the same (MetaFrame) counters that I mentioned in an above post, so I need to correct myself on that point since, by implication, Systrack can also access these metrics.

Maybe it's time for an FAQ or two on System Monitoring principles and practices - especially from those who have had successful results using particular tools ;-)

Final note: If the farm is growing into double figures of servers, it may well be worth considering a move to XPa for Load Management.

IntrepidUsa - it'd be great to know how successful you were in diagnosing your issues.

CitrixEngineer@yahoo.co.uk