Is XenDesktop the right tool for the job?

So after the storm announced yesterday, people all over the world started complaining about the new licensing that came out with XenDesktop 4. This also prompted a heated discussion within the CTP community, with some explaining why they were frustrated with such licensing changes.

For sure I can understand the issue for certain customers. For example, for some educational institutions with 25,000+ users but willing to deploy a VDI solution using XenDesktop 4 for only 1,000 people, what do they do? If it is licensed per user, how does this work now? Do they need to pay for 25,000+ licenses?

Using the old Concurrent User licensing, all they needed was 1,000 licenses. And that is where the discussion with the others CTP started from my standpoint. 🙂

I am posting here exactly what I told all the other CTPs. First of all I think Citrix wants to change the idea that XenDesktop is a remote access solution. They want to make the market aware XenDesktop is actually a desktop replacement solution, going head-to-head with what we call today ‘physical desktop’. In that sense, if you have 1000 PCs in your office and you want to go VDI you need 1000 hosted desktops. So if we do see XenDesktop as a replacement for the physical desktop, 1:1 licensing makes sense. And again, from a marketing perspective it makes sense if you want to get rid of the old stigma of being a ‘Remote Access’ solution.

If you have 10,000 users and you want to have only 100 connecting remotely to your hosted solution, it is clear to me you are really trying to deploy a remote access solution and for that, there is XenApp. Many, including my friend and fellow CTP Joe Shonk, argued XenDesktop brings simplicity to the table as you do not have to worry about your apps, what you do when using XenApp. Well my take on this is simple and I ask you to provide me feedback on this in the comments area.

As of today, at least for my customers (note I have been doing TS/Citrix since they were born as products, so at least 10 years), in 99% of the cases we were able to make the applications work. Sure, some we had to tweak but the bottom line is they worked. And as of today, with application virtualization, there is a chance you will get the ones that did not work, working. So the whole “XenDesktop is app friendly and XenApp is app hater” does not cut for me. Add to that the fact that several XenDesktop customers are using Citrix own app virtualization/streaming solution to package and deploy the apps so in this case, that exact same app could be used on XenApp, making it not better or worse than XenDesktop.

Even though using XenDesktop as a remote access solution works and works properly, I see this as using the wrong tool for the job in most cases. I do know I can load a web server on my iPhone but should I use it to then host my website? You get the idea.

I am aware in some very special cases XenApp may indeed not work and XenDesktop or similar alternatives would be the only way to go. If you are on that boat, good for you.

But for most cases, nope. XenApp delivers it. And it has concurrent licensing. 🙂

The good news is Citrix is listening and is dealing with such cases, people that are trying to use XenDesktop in weird ways (sorry I could not resist), on a case-by-case basis.

I also heard on the CTP program they are releasing a new version called XenRAS that is pretty much XenDesktop but tweaked to work as a RAS solution, only accepting dial-up connections and with concurrent licensing. Official announcement should come soon from Citrix.

Oh crap. That was under NDA.


3,123 total views, 2 views today

7 Replies to “Is XenDesktop the right tool for the job?”

  1. Thing is, the way we want to use XenDesktop isn’t weird at all. We want to use it to do OS Streaming to real PCs for our labs, not as a remote access solution and really not even as a “virtual desktop” solution.

    OS Streaming is one of the premium features in XD, based on it not existing in the Standard edition. And that’s cool – it is a valuable feature.

    Last week, our XD rollout would have cost $195,000 max at retail. Probably half that at our contract price, and heck, maybe even less at that volume.

    This week that same rollout would cost $6,300,000 for our 28,000 users, maybe half that once the contract is updated (only $3 million, wow), and hopefully less at that volume.

    Browse through some of those customer videos Citrix has on its website sometime. Notice how many of those are school systems or universities. Some of those are doing VDI, but a lot of them are doing a whole lot of OS Streaming. Whether it is called labs or classrooms or whatever, there is a huge market out there just in education (not to mention other industries with dense user to device ratios over multiple shifts) that would be eliminated by per-user pricing.

    I’ve seen some comments that indicate that Citrix has always been expensive, so we should just suck it up and pay it if we want the best product. And the thing is, that made sense last week. I have had to bust my butt making the value argument, and I’ve succeeded. But that was at last week’s CCU pricing. I couldn’t sell value at per-user pricing if I were Joe Isuzu – when you’re talking about user populations like ours and use cases like ours the price differential isn’t something you can argue – you can only laugh about it.

    Before I started seeing some folks talk about Citrix being willing to consider alternative models for education, and before I heard from the VP over XenDesktop after I commented on his blog post, I’d explained the licensing change to some of folks on our project team that is working on the Citrix rollout. To a person they all stood there, stunned and the first words out of their mouths were, “Well, I guess we have to find another solution, or keep doing things the way we’re doing them now.”

    I’m believe Citrix will see offer a workable licensing program for education, mainly because Sumit Dhawan emailed me and said they were, and that my labs use case is his #1 priority to address. I *really* want to implement XenDesktop – and I think Cirix will recognize they really want us to be able to, so I’m hopeful.

    Oh, and as for XenApp being a workable alternative – in many ways it is. Unfortunately, XenApp alone does not give me the ability to run our labs using real PCs and a single image the way XenDesktop does. Even if I stripped our image down to XP, browsers, plugins, and the Citrix client and fed all of our apps (tons of them) to it via XenApp, I’d still have a thousand XP machines out there having to be managed in the same old much less efficient manner. Maybe if someone would give me a thousand XPe thin clients I could switch to a XenApp only model, but I have to work with the equipment I have, and the best way to do that seems like XenDesktop.

    Anyway, I really appreciate you sharing your thoughts about this and all of the discussion going on.

    1. Agreed Mike. The ‘weird’ comment was just a joke. 🙂
      What I see is certain environments will try to use XenDesktop in a way that conflicts with this ‘push’ by Citrix, targetting physical desktops and that is where the problem lies.
      But as I mentioned they are now dealing with such scenarios on a per case basis. I am sure you will be covered. I am sending your contact info directly to them so you guys get this sorted out.

  2. You make VERY valid points and I don’t disagree completely with you. I absolutely agree that I think Citrix is trying to get people into the mindset of XD being a desktop replacement.

    That being said, take your statement from above:
    “First of all I think Citrix wants to change the idea that XenDesktop is a remote access solution.”

    This goes against what you stated in your previous blog post:
    “Citrix says when accessing XenDesktop it means the user will have access to his desktop 24/7 while on XenApp he accesses the apps on a needed basis, not necessarily 24/7 so concurrent user makes sense for XenApp but not XenDesktop.”

    So which is it? If their justification is that a user will/can access his/her desktop 24/7 I see that as it being a remote access solution. Yes, you have the same capability today by RDPing into your physical machine [after launching MSTSC as a published app of course 😉 ] but it is still remote access.

    My biggest concern with the license change is that I see VDI in a position to handle the exact kind of scenario Mark listed above and then some. Labs/Training rooms, Offshore workers, shift workers. When you look at each of those cases you see a common pattern forming. Standardized desktops that are locked down and are likely NOT going to be used 24/7 by the same person. Since these are the most likely fit for VDI at the moment per user licensing kills the adoption and push to improve the tech. The argument that these could be hosted on XenApp with a desktop is not the point since the end goal is to give the user the best experience possible.

    In no way, shape or form do I think VDI is ready to replace everyone’s physical machine..The technology, stability, and feature set is just not there yet. Things like XenClient will make things better though and it is a step in the right direction.

    Hopefully this made sense but its late and I am tired….

    1. Oh my first post on the subject, the XenDesktop 4 announcement one, I was just quoting what Citrix was telling us CTPs in the conference call. It does not reflect my opinion in the subject at all. That I saved for the second post, this one you just wrote your comment.
      I will have another post today on the topic as at the end, to me, if they do want to push it as a physical desktop replacement, per device licensing is what makes sense and not per user. Subject for another post. 🙂

  3. As usual Claudio I mostly agree with you. The only difference I have is that we still come across apps that don’t play well in XenApp. For years it was for mostly technical reasons and we pulled off our share of miracles to make these work (and I go back to when Citrix was Winview on OS/2!). We have used SoftGrid since it’s inception and a range of other tools and self built best practises, yet, we still do have apps reqularly that are problematice.

    Another driver is user experience, for example AutoCad and Adobe CS suite simply work better for users when integrated into a standard desktop environment. Sure you CAN make it work on TS, but then you leave the customer with the daunting task of maintenance, upgrades and support. We push some of these types of apps to VDI are for support reasons, i.e. the vendor does not support a server OS and the customer wants to be on a supported platform!!

    So I still think that VDI has an important compatibility play despite the greatly improved situation with app compat on XenAPP

    Steve Greenberg, CTP
    Thin Client Computing

  4. Steve, exactly for apps that don’t play well on TS Citrix just brought out the VM hosted app feature of XenApp FP2, and with HDX 3D we should be able to push out even these graphic intense apps.

    That is why we thing the new XenDesktop 4 announcement fits perfectly, as it ties it all together.

    Markus Klein
    Citrix Systems Central Europe

    1. That is it. I see XenDesktop and XenApp as very close buddies. I would say as of today, considering the costs associated with VDI and the fact it is still in its infancy as a technology/industry, the perfect solution is a mix of SBC and VDI. SBC has been around forever and even from a resource standpoint it is much easier to find someone that is great with SBC than it is with VDI. So ideally you would use XenApp for probably 95% of your apps and for these 5% that do not work on it, give it to the users, seamlessly, through XenDesktop hosted apps. Simple as that. And transparent to the users.
      Honestly I do not see XenDesktop or any other VDI solution for that matter, mature enough and more than that, priced attractively enough, to become a desktop replacement in the short term. That is why at this stage per user licensing makes no sense.
      Down the road as people move towards a VDI model (and trust me, XenClient will play a HUGE part here once it is part of the whole XenDesktop – what may happen one day, who knows) than per device licensing is probably the way to go.
      But today, in a mixed XenApp/XenDesktop world as you just mentioned, per CCU is the only way to go.

Leave a Reply

Your email address will not be published. Required fields are marked *