05 Mar 2008 05:44 PM EST
[ Tags: unix,  xenapp for unix,  linux ]

As a follow up to Carlo's post on XenApp for UNIX, I would like to discuss our future for the UNIX product. XenApp for UNIX is a fully supported, maintained and enhanced product. Since we released Presentation Server for UNIX 4.0, the product has been following an incremental feature delivery model. Since the 2005 release we have added over 80 feature enhancements like seamless improvements, session query utility, enhanced diagnostic logging, roaming user support, adding support  for Solaris x86/x64 platform, Solaris SPARC license server, Virtual Channel SDK, Enhanced keyboard and wheel mouse support, Solaris zones support, enhanced server farm publishing options etc. Instead of coming up with a brand new release (like PS for UNIX 4.5 or 5.0), we have opted to get these enhancements as public hot fixes and feature packs. e.g. we added Solaris x86/x64 support when we released PS 4.5 Feature Pack 1. And we will have the next feature pack update for UNIX that will align with the upcoming Delaware release.

The reason for using this delivery model is it speeds up our feature development and helps our customers easily adopt the functionality they need. The customer can install these updates as either hot fixes or as feature packs based on their needs. Of course, you need to be current on SA in order to use the features.

Regarding support for Linux platform, we still don't see a huge market for Linux apps. Also, we might not have native Linux support but some of our customers use XenApp for UNIX as a proxy to serve Linux applications. We will soon have a KB article explaining how you can do that.

Permalink | Comments (6) |

Thanks for responding to my questions Sridhar, that clears up the concerns I had with the future of XenApp on *NIX. The justification for Linux makes sense since I've done the same to deliver OpenVMS apps via X over the years and its always worked great.

Oil and Gas is a big sector with plenty of money to throw your way if you can help us virtualise geoscience apps running on Linux.

Posted by Anonymous at Apr 03, 2008 04:08 | Reply To This

I fail to see how asking customers to double their hardware to support Linux makes any sense. If people are using XAU as a proxy to Linux systems then wouldn't those customers benefit even more from a native Linux solution? What is so hard about supporting Red Hat? They have a very, very nice patching and build system (their Satellite Server technology) which is something we find very, very attractive -- especially now that PCI DSS compliance has become a high priority. I am very disappointed that you continue to shun Linux the way you do and it seems disingenuous to say there is little interest while more and more companies are dropping UNIX in favor of Linux. As an example OPNET just dropped Solaris support. And Solaris Containers for Linux does not look like it is ready for production yet and I cannot see vendors supporting that environment even if it was.

In XAU the load on a server is normally made up of the XAU itself and the applications being remotely served so by taking application load off the XAU server it leaves capacity in the XAU server to handle more sessions. So in a multi server farm situation there is unlikely to be a doubling up on actual hardware. Of course exact hardware requirements is very much dependent on the type of deployment and in a single server deployment doubling up is the likely outcome. Of course, if the servers can be virtualised and the hardware has the capacity, then no additional hardware may be required.

The question on supporting a Linux version of XAU is not a technical one in so much as building a version that runs on some Linux distributions has been achieved in our development labs. The questions that need answering are the following.

  • What Linux distributions should be supported as release testing on many distributions that themselves have multiple versions is cost prohibitive.  Support is much more than just having binaries that appear to run correctly.
  • What demand there is for a Linux version is another important question to have answered. Not all applications that migrate to Linux have a need for a XAU solution, only those that have an interface to be delivered to end users fit the category. I believe Linux has such a user base in the Electronic design market and Geosciences market where there are ISV supported applications. I also believe the Banking industry have bespoke solutions also running on Linux. What has been hard to ascertain is if this market can support the investment required for us to release a solution.

To help us answer these we always advocate customers raise enhancement requests but perhaps something more lightweight, say a Blog poll can be set up as well.

Thanks to PCI DSS package management has become muc more important to us and Red Hat's satellite server technology is very attractive. Right now we have a lot of compiled apps like KDE on Solaris that if we switched to Red Hat we could just use their packages and thus greatly reduce time spent recompiling to apply new patches. If we could get a Red Hat version as part of our Subscription Advantage we would switch to it and likely stop evaluating alternatives that do run on Linux. It's mind boggling that HP-UX is still supported but Linux is not. Are you saying that you conceding the Linux market to Secure Global Desktop and NOMACHINE? That seems to be the message. As for which Linux distros -- Red Hat seems a safe choice (just try and stay current).  Note: we only have Subscription Advantage, so we cannot raise an enhancement request.

From a distro perspective Red Hat is no longer the only viable production game in the town, the Ubuntu 8.04 LTS server distro is a solid debian based solution for a lot of people.  I personally find dealing with package management/upgrades easier with .dev files than .rpm files.  To be honest I think both should be supported.

To give you an example, one of the most frustrating things that I have with Citrix as a ubuntu user is not that the only download available is in .rpm format, I can convert that, but rather that it is only available for a 2.4 kernal.  Who even uses 2.4 any more? Everyone is 2.6+

The great thing about linux/unix people is if you release the source for a 2.6 kernel net6vpn the community will get it packaged for you.  Just some thoughts.