18 Jul 2008 04:39 PM EDT

posted in XenApp by Barry Flanagan

If you could wave a magic wand and have any one single feature in the next release of Citrix XenApp, what would it be?

While XenApp has literally hundreds of features that have been added over the last 10 years as the product has evolved from MetaFrame 1.0,1.8, XP, FR1 though FR3, 3.0, then 4.0 and now 4.5, is there one feature you really want to have but have not seen yet?

I pulled in a few ideas I have received into a poll. If you would like to add others to the list, post them in the comments and I will add your suggestion. I am looking for big home run features, but ideas that help you in your day to day job are fine as well.

From my past experience, many of you just do not have the time to keep up with every feature added to XenApp/CPS over time, especially if you are migrating to every other release. I have a theory about that as well (but I am saving that theory for a later post). It will be interesting to see if any feature ideas are submitted that have already been added in a past release.

Of course, there is no guarantee anything on this list can or will be included in the future (since I am not on the XenApp product team). I do guarantee I will communicate the results of this poll to that team.

This is a very preliminary list based on an informal survey I took recently. Instead of editing it, I am just posting it to get the discussion (and voting) started. If you want to add a feature to this list, post it in the comments and I will add it. Focus on the problem you need to solve.

If you could wave a magic wand, what feature would you put into the next release of XenApp? Choose
SpeedScreen Multimedia support across all client platforms
Enhanced smart phone integration
Enhanced Mac integration
Live migration of individual sessions
Diff tool for Microsoft and Citrix patch levels on all servers
Integration with third party app virtualization solutions (MAV or SVS)
Greatly enhanced scalability
Security Analysis and Lockdown Tool
Pool of Pre-Launched Sessions for Faster Launching
Management through an Active Directory OU



UPDATE:Two Additional Choices offered based on comments.

If there is a lot of interest in this poll, I will post follow ups on the features requested and similar polls for other products. My goal here is to get unfiltered feedback from you about what you want to see in the product and how we can improve the product to solve the problems you face.

UPDATE:The response to date on this poll has been excellent. Votes are still coming into the poll. I have sent a screen shot of the results to date to a member fo the XenApp team to ensure the product team is aware of these requests. I am working to get some members of that team to discuss of these requests and the comments posted here.

Permalink | Comments (28) |

Comments  (Hide Comments)

I voted for 'Live migration of individual sessions'.  I kow there's very little chance of this ever being implemented, but the down-stream benefits it would bring be substantial.  If I can't have that the next item on the list would be 'Security Analysis and Lockdown Tool'.

I don't understand why scalability enhancements is on the list - I'm running 65,000 concurrent sessions today without any significant issues and don't have any concerns about pushing this number well past 100,000, and if the poll is asking if I want per server scalability improvments the the anyser is a firm no, the last thing I want is more eggs in the one basket.

Other features that I really want to see are:

Faster session startup and application launch times (possibly implemented by having a pool of pre-launched disconnected sessions with the app already running ... )

The ability to manage groups of servers (publish apps, install packages, assign load evaluators etc) where the group is defined as an AD OU. 

Integration with Provisioning Server/XenServer to build servers on demand in response to changes in demand/capacity ( and to power them off again when no longer needed).  Or even better create an open interface to allow 3rd party provisioning tools to access application demand/capacity data.

Posted by Simon Bramfitt at Jul 19, 2008 12:59 | Reply To This

Simon,

Thanks for your comments and suggestions. I added your first two suggestions. The auto-provisioning suggestion should be possible once WorkFlow Studio is released. I will ensure that is the case.

I can certainly understand your opposition to greater single server scalability. Many people share your concern about too many eggs in one basket. Many others need greatly enhanced single server scalability because of lack of resources like data center space. Numerous large enterprises pay large system integrators a per month per server maintenance charge, so enhanced scalability is very important to help control monthly costs. As with any other feature, some may benefit a great deal from such a feature while others may not find any value in it.

Posted by Barry Flanagan at Jul 20, 2008 22:54Updated by Barry Flanagan | Reply To This

The scale-up capabilities with 64bit servers has been a very welcome addition to my environment. I have purchased nothing but 64bit servers for the past two years and have the goal of zero 32bit physical servers in the future. A company that is handling 65000 concurrent sessions typically is not running very many applications and will kill many users on a larger server. I look at it a little differently and we know that it is all about the applications that we are hosting. My goal is to run as many applications on each server as possible and I have approximately 500. You might still be dropping 200 users when you have a problem, but these users are not all on one application and the pain is spread across multiple business units.

BTW, I voted for the live migration of individual sessions. In many ways I think of this as the next logical progression for abstraction. First came the OS, then the applications, why not the profiles and user sessions? I do understand the complexity of this ask, but if it was easy it would already be done!

Posted by Jonathan McCully at Jul 22, 2008 09:34 | Reply To This

On single server scalability

I've no real issues with a general trend towards more sessions per server, I just don't think that Citrix should be assigning resources to develop a solution here. The availability of 64-bit Windows, multi-core processors, virtualization and higher density blade systems has greatly increased the number of sessions that it is is possible to run per processor/server/rack already and there's no reason to suggest that this trend will be arrested any time soon.  I would have thought that anyone who is seriously impacted by per server scalability issues can tackle that problem far more readily by forklifting in new hardware than by deploying a new version of XenApp.

On session migration

There's already been a lot of input to why this is a desirable capability, so I'll limit my comments to expanding one one point that was already made my somone else.

Aside from the availability and administration benefits; session migration would provide another tool in the capacity/demand space.  If I plot utilization in our environment I see a typical bell curve with the peak (65 K sessions) occuring at about 11:30 am and the low point (3 K sessions) occuring at about 3 am.  At the low point I only need about 50 servers to host all the active sessions, but due to the way sessions are distributed across the entire environment I'm running many hundreds of servers with each hosing only a couple of sessions.  Session migration would be a highly effective way to allow me to concentrate sessions onto the smallest number of servers needed to meet demand, run those servers at maximum efficiency, and power down the rest. 

The straight financial benefits of doing this are so big that having this capability would be the easiest sale any vendor ever made.

On provisioning

I've high hopes for WorkFlow Studio, using it to provide the glue to link XenApp capacity/demand managment to Provisioning Server/XenServer could well be a game changing event.  I've been fighting with this problem for a little while now and keep banging my head aginst the lack of a simple means of determining application capacity within a XenApp farm, if WFS can do that I'll be very happy satisfied for now.

Posted by Simon Bramfitt at Jul 22, 2008 14:24 | Reply To This

65.000 cc sessions? WOW!

Posted by Anonymous at Jul 19, 2008 16:20 | Reply To This

I agree to Simon, Live migration of individual sessions would be a huge benefit in operations. This could make the user experience a lot more reliable. e.g. if the performance monitoring detects that a single CPS is going to run into a performance bottleneck, then all user session on the CPS could just be migrated to other servers.

Posted by Christoph Wegener at Jul 20, 2008 19:28 | Reply To This

Thanks for your comments Christopher.

Posted by Barry Flanagan at Jul 20, 2008 22:55 | Reply To This

For this specific concern (server bottlenecks) wouldn't an entire virtualized XenApp/CPS server running in a VM allow you to deal with this issue through a live migration? With XenApp on xenServer you could migrate the entire VM to another server. Would that work for this instance?

Posted by Barry Flanagan at Jul 20, 2008 23:50 | Reply To This

I also agree with other comments above that "Live migration of individual sessions" would make my life as an Citrix admin soooooo much easier.  In a perfect world I'd love to be able to implement a process where I could power down a silo of under-utilized servers overnight when we don't have anyone in the office.  The problem we have at the moment in dealing with those open sessions on the servers.

Posted by Anonymous at Jul 20, 2008 20:26 | Reply To This

Thanks for the feedback.

Posted by Barry Flanagan at Jul 20, 2008 22:56 | Reply To This

Since live migration of specific sessions is the leading vote getter to date, I woudl love to hear more comments about this specific goal. What issues lead you to vote for this specific feature (other than the already mentioned server bottlenecks)?

I wonder if XenApp/CPS running on XenServer would alleviate the concern for most issue? If not, how about XenApp running on XenServer and protected by Marathon's everRun product? Marathon can duplicate either an entire physical server running XenApp/CPS and balance the load between the two of them (which appear as one server to all clients) or the product can virtualize a VM of XenApp/CPS between two physical hosts. I am not sure if that alleviates the concerns of an OS failure on the XenApp/CPS server, but I will ask. If it does (meaning the Marathon everRun server can continue to run even if one of the hosts blue screens) would that provide the availability you need?

I recognize there is additional costs involved. Just for the sake of this discussion, I would like to focus on what problems you need to solve and figure out if it is possible to solve them now.

Posted by Barry Flanagan at Jul 20, 2008 23:58Updated by Barry Flanagan | Reply To This

I had a discussion with Michael Bilancieri(Director of Products) at Marathon Technologies.

The Marathon everRun product for physical servers includes system level fault tolerance. This means that everRun can provide fault tolerance in some scenarios where there is an OS failure.

Marathon everRun FT

Here is a quote form the description from Michael -

"With our existing FT product providing system-level FT for physical Windows servers, the answer is somewhat dependent on the cause of the bluescreen. Our system-level solution actually runs two separate instances of the OS and application on two separate servers in complete lockstep, meaning that all instructions are being performed synchronously on two separate servers. If the bluescreen was caused by an application or OS software bug (divide by 0, for example), since both sides are executing the same instructions at the same time, both would have the failure. However if the bluescreen was caused by flaky hardware (bad RAM, etc), only that side would go down and the application would continue to execute on the other system without any interruption."

Gabe Carrejo of Citrix did some videos of his testing with Marathon everRun product and XenApp. I blogged about this testing and included several videos Gabe made here.

Marathon does not yet have system level fault tolerance built into its everRun VM product, but that is coming soon.

I understand this is something you would like to see specifically from Citrix. At least some sort of solution is available to today.

Posted by Barry Flanagan at Jul 21, 2008 13:02 | Reply To This

Barry

Clearly Marathon have a compelling solution for people with availability or business continuity problems to solve, but we can't really consider everRun as a means of addressing the operational issues identifed here.  VDI solutions that can provide the level of flexibility needed do already exist, but unfortunately VDI does not achieve anything like the level of efficiency (number of sessions per server) that WTS + XA can deliver, so can't be considered a realistic means of addressing this problem. So really we are no further forward.

I wonder if given the level of interest already expressed, it may be worth asking the community just how valuable this capability really is and attempt to firm up some use cases and requirements...

Posted by Simon Bramfitt at Jul 25, 2008 13:03 | Reply To This

Simon,

THanks for the feedback. I am working on getting one of our high level developers or architects to provide some feedback form their perspective on this feature. I am going to ask our Microsoft relationship team if I can interview one of the Terminal Services architects regarding this as well. Hopefully I will have more info on this soon.

Posted by Barry Flanagan at Jul 29, 2008 09:45 | Reply To This

Live migration of individual sessions by far

Posted by Anonymous at Jul 21, 2008 07:52 | Reply To This

Barry,

I think you're missing the reason for people wanting LiveMigration of sessions between CPS servers. Wanting to move users from one server to another is sometimes due to a server having a hardware related issue (pre-failure notification of memory module, needing to add memory/CPU, etc). However, those issues are few and far between versus the much more common issue of needing to perform some type of OS maintenance whether it be applying a hotfix, making a machine change that requires a reboot, etc). These are the situations that occur most frequently and are the ones that people want the session portability for. Personally I think session portability is a great concept, but I don't think it'll work quite like everyone hopes for.

Shawn

http://www.shawnbass.com 

Posted by Shawn Bass at Jul 21, 2008 10:54 | Reply To This

Shawn,

Thanks for jumping in on the conversation.

I understand that OS issues are a big driver in this. That is why I wrote above "I am not sure if that alleviates the concerns of an OS failure on the XenApp/CPS server, but I will ask". I am not sure if the everRun technology from Marathon could deal with a bluescreen or need to apply a patch to one side of the cluster. I am going to contact Marathon today and find out. I will post an answer here once I get it.

I am curious is this concern is by far the biggest driver in the desire for this feature, or if there are others. As you know, this type of feature would be very difficult to provide. But I wonder if there are other ways to solve the issue.

Posted by Barry Flanagan at Jul 21, 2008 11:05 | Reply To This

Completely agree with Shawn, the ability to put a server(s) into a maintenance mode that then automatically moves live sessions to another server for among other things "Patch Tuesday" would be a great help for administrators and less disruption for users at the same time. If it could also do this procedure in a load-balancing scenario, even better.

Posted by Neil Hiorns at Jul 21, 2008 11:31 | Reply To This

The way i see this feature it is a natural next step in demand from us to you. We are now moving guests around with Motion. this allows us to service hardware without(or with less) downtime. With this strong tool at hand i think the natural next step on the wishlist would have to be user-migration, to be able to manage the OS on the servers without having to book a service window 3 weeks in advance or similar.

So.. I think the reason for this sudden desire is simply and mainly that we are all getting used to providing almost zero downtime, while being able to do the administrative tasks during the day. Hardware is cheap, so redundant boxes are easily bought, however man/brainpower is alot more expensive to come by. .. and on top of this all of the sudden people want to have less than 60 hours a week, holidays _every_ year.. 

/René Vester

Posted by Rene Vester at Jul 22, 2008 06:59 | Reply To This

A simple item (in my mind anyway) is that I would like to see realtime QFARM reporting area within the Management Consoles rather than scripting. I realize the amount of different items that one can generate using QFARM is large but a few simple ones such as reporting a full load would be nice. Especially when delegating duties to a help desk.

Posted by Trevor Svienson at Jul 21, 2008 12:42 | Reply To This

Collapse all the management consoles into the AMC.

Posted by Anonymous at Jul 21, 2008 15:00 | Reply To This

How about a built-in server cloning utility (perhaps a cloning prep and post-cloning utility), and a built-in User-Profile Manager?

Posted by Norvin Llorando at Jul 22, 2008 01:33 | Reply To This

In the Access console - Servers List, A "list all servers "  feature , rather then having to trawl thru folders,
Citrix reboot, have an editable variable "disable logons for *** hours prior to reboot ...  curretly max delay is 60 mins , how about 2 \ 4 \ 8 12 \ 24 hours please!!!!
reintroduce "pause" in CSG service, & fire the person that took this away ....

Posted by Anonymous at Jul 22, 2008 07:12 | Reply To This

I agree, extending the "Disable logons before reboot at:" would be very useful and it should also have the "Send message" with the same extended times as well. Moving it to AMC would be even better, but as a quick fix could live with it still being in the PSC.

Posted by Neil Hiorns at Jul 22, 2008 07:32 | Reply To This

How about the ability to create ICA files from the management console again?

Posted by Anonymous at Jul 23, 2008 17:42 | Reply To This

Why?

Posted by Anonymous at Jul 23, 2008 20:39 | Reply To This

I posted a follow up to this article here.

This is the first of several follow ups I plan to do.

Posted by Barry Flanagan at Jul 29, 2008 15:56 | Reply To This

I would like to change the way session timeouts works.

How about session time out based on users, groups or even at home vs at the office.

 I would like no time out when I work at the office, but if I reconnect to my session at home (acces through CAG), I would like to have idel time outs. This is a security issue at my compay

Posted by Anonymous at Aug 15, 2008 06:51 | Reply To This