Before jumping to the conclusion this is one of my standard knee-jerk rants full of bad language and poor grammar, it's not meant to be. Although I don't expect the CEO of Citrix to know anything about my alias or to read this, I do feel this is an important topic of concern and hope at least some people will understand the spirit of what I am trying to say.
Several years ago, Brian Madden asked the question Can Citrix become a thought leader again? How did Citrix respond? Well today Citrix has a vibrant blog with all levels of the organization contributing, the Access message has significantly evolved, and there have been many positive efforts engaging with customers. Additionally Citrix has ramped up their talent with some great additions in all areas who regularly reach out to the community. So for that I say thank you and well done.
So what is the concern?
Desktop virtualization is here. VMware View is not relevant to enterprise customers for desktop virtualization for many reasons that I'm not going to dig into today blog but clearly Citrix knows this. As a result many of us are trying to implement desktop virtualization today using the Citrix stack, even when it's on top of an ESX hypervisor. As that happens, Citix is becoming more strategic to organizations, and that means Citrx has to continue it's evolution and become an enterprise class provider. To do that several areas need to be addressed:
- Product leadership
- TCO & other items
If Citrix releases a product, as a customer I need to be able to trust the quality of what I'm getting. Sadly that has not been the case with XenDesktop. I've sat back and watched under the premise that the product was new and that rapid innovation was occurring, so therefore problems would occur. However we're now at XenDesktop 4.x and still basic quality issues are rampant in the product. The XenApp product is better, granted it's more mature, but even there it needs to be much better. If you don't believe me, go and look at all the XenDesktop client connectivity, a.k.a. "VDA problems" that have plagued the product for so long. It's clear that Citrix does a sub-par job of testing their XenDesktop product. As a result I (and I'm sure many others) have gotten lots of black eyes with from my user base due to Citrix quality. It feels like Citrix expects their customers to be their testers, and ultimately it results in very slow production deployments of the Citrix infrastructure which ultimately means one can't take advantage of new features quickly. (And is it reasonable to charge maintenance for a product I can't deploy until many months of basic bug resolution that should have been caught by Citrix in the first place?) Of course this is just one example. I could also mention the basic lack of information on how the whole Citrix stack integrates between all the various moving parts: which versions work with which, what's been tested, etc.
Frankly I have to question the competence of the product and engineering teams that develop XenDesktop. Of course there are bugs, but the efforts to date have been rather amateur, despite great response from support teams after the fact. If Citrix wants to drive broader adoption of their stack, then this level of quality can't continue. As a contrast, when I get a product form VMware, I have a much higher level of confidence that it will work and I will be able to deploy it into production faster.
While Citrix has done a great job of opening up in the last few years with blogs and community efforts, current XenDesktop product leadership is bringing back a lot of the old Citrix mindset that we didn't like back. (e.g. Very closed and false certitude that their way is best.) That is how Microsoft behaves, but in this critical time in the evolution and growth of Citrix, the Citrix leadership can't afford to be this way as they need to win customer trust. Interactions specifically with the XenDesktop product leadership have mostly resulted in nothing but canned answers and excuses and very little constructive two-way dialogue that results in action. (Although the recent the addition of the delegated administration feature provides a ray of hope.)
The quality discussions have been there from the beginning, but there are fundamental things that are totally ignored which are important to those of us implementing, not some figment of the imagination of some manager who knows nothing about the desktop and fantasizes about mass market needs that may never happen. It is we, the early customers who have put our faith in Citrix, taken big risks in implementation every day, and suffered black eyes due to lack of quality, and yet Citrix product teams tell us again and again that certain things are just not needed, despite the obvious need to those of us who understand the desktop.
A good example is Reverse Seamless which has been talked about for years. Every conversation with the XenDesktop team has resulted in excuses. Why can't Citrix give a clear answer on this? Are they ever going to do it? It's needed by many of us. Sure there's a commercial version of this feature now from RES Software, but this is irrelevant as I'm not going to pay for something that needs to be a native feature of XenDesktop.
Here's another example where there is still no direction from Citrix and just excuses from the XenDesktop team: HDX Connect. In the comments section Shawn Bass highlights key use cases for implementation in the real world: (For the record I haven't asked to or corresponded with Shawn while writing this blog article.)
"There's another use case that wasn't covered here. That of the IT person who's responsible for supporting/troubleshooting XenDesktop. Today, when there's issues with broker registration, firewall rules, etc. there isn't a way to confirm the basic ICA Client -> ICA Host connectivity is working without incident. This is because the ICA/CGP ports are not opened until after the broker tells the VDA to do so. This makes troubleshooting ICA/CGP connectivity almost impossible. If this was opened up, one could just fire up an ICA client, type in a machine name and boom you're in. Combine this with the use cases of:
- Immediate DR ability without having to assign a bunch of machine names to people (I know there's pools, but if the app set doesn't match then this is still a problem)
- Being able to have a single user use multiple XenDesktop systems without having to create multiple desktop groups.
And there's probably other use cases I'm not thinking of right now. But the biggest one for me is enabling troubleshooting and firewall traversal."
A key point Shawn makes is about a way to just connect without a broker and no management for troubleshooting purposes. It's an enterprise need. Due to all the quality issues with the broker as well, it's even more important. Please tell us, why have the XenDesktop team completely dodged the issue with HDX Connect? Where is it? What is the direction? Why is it so hard, since we all know you can give it to people given the work done with IBM and Kaviza who can use HDX in their stack.
I'll end this section by simply saying that this kind of ignorance can't continue. The current XenDesktop leadership has done little to gain trust. I have little faith in them driving pragmatic customer needs moving forward. Bring back the more open Citrix who listens. Please have product leaders who "get" the desktop, not the puppets put forward today that ignore customers who are currently implementing by talking about broader market fantasy that may never come true.
TCO & other items
I'm also tired of hearing about VMware Project Horizon delusion and futures that are many years away and a total lack of focus on the real world of the next three years. TCO is still a big issue for broad desktop virtualization. VDI costs more that PCs although it offers other benefits that may make it worth it to drive business needs as in my case. XenApp is the cheapest model, which is good, but I'm also tired of hearing debates of VDI vs. RDS/XenApp. So here are some suggestions:
- VDI/XenApp hosted desktops can be made cheaper with better personalization. Brian has a good blog on this. The bottom line is "how to move people beyond persistent desktops?" is a piece of the puzzle that needs to be solved. The current product capabilities are not there, and should be evolved to help both models. Citrix's own CTO Harry Labana articulated the problem quiet well here, so where are the results?
- Storage for the VDI model is too expensive. Local cheap commodity storage is the future, and another good Brian blog here. Citrix should take a leadership position here and help guide the market. Ignore all the fools who invest in SAN for the desktop to store stateless system files.
- XenApp. It's unclear to the world what the future of XenApp is for the published desktop model, (which is still the lowest cost deployment model for many). XenApp 6 is pretty good, but it's x64 only. What about 32-bit desktops? Will there be innovation in both models?
- Citrix should help fix the problems with App-V (there are many), assuming Microsoft will let them. If Microsoft wants to continue to not innovate with App-V I would strongly encourage Citrix to innovate in application virtualization, just as VMware is doing with ThinApp and their IE6 solution (which again many of us need to run locally in our environments).
- Integration. If Citrix truly believes that VDI and XenDesktop will be part of the future, then they need to demonstrate that commitment by bringing the products together. Citrix marketing talks about a Flexcast model of multiple desktops which is fine. But the reality is it's several products bundled together to tell a bigger story. That's also fine, but we need to see a simpler infrastructure in the future that brings this stuff together.This will result in a very powerful product. Quest vWorkspace has some of this today, although you will have to configure all their integrated solutions which adds back to the complexity.)
- XenClient. Great! It's released! But it's slow, the user experience with multiple VMs needs lots of innovation, and there is a weak management story. As I've written on the comments on this blog many times, MokaFive still has the best enterprise management story for client hypervisor use cases. Then there's that Apple thing also, so having a rich-managed Type 2 solution for Macs to start with would be an awesome addition to the Citrix story. The bottom line here is that Type 1 will be slow to evolve, and offline Apple use cases are a problem today. What can Citrix do to help?
A critical time in the history of Citrix
Desktop virtualization and its evolution are very important for those us already on this path who have invested a lot of skin in the game. There is a huge opportunity ahead for Citrix. VMware is faltering all the way, and doesn't understand the desktop. Citrix has made great progress, has some great assets and some great people that have have positioned them well. But to adopt desktop virtualization, Citrix needs to be a trusted critical part of a customer's enterprise infrastructure. It needs to be enterprise class.
Therefore product leadership that is able to deliver quality products on time, build features that aid real implementation, listen and build trust by not ignoring obvious requests as opposed to arrogant "I know better answers" will be key to ongoing success. Citrix is not Microsoft and therefore can't afford to behave like them. Citrix requires leadership who get it. (Granted there are some who do get it at Citrix.) I've alway enjoyed listening to Mark Templeton whom I think has great vision and comes across very sincerely. I am sure Mark and team will (as usual) paint a compelling vision at Synergy next week which I will follow. However given my experiences over the last few years with XenDesktop, I'm wondering how much I can count on the execution and leadership charted with delivering enterprise class desktop virtualization? No matter how good a vision is painted at Synergy, unless I can trust Citrix to deliver reliably in the enterprise today, that vision will quickly become delusion.