XenDesktop – some notes, post iForum
I’ve had a day at iForum and feel ready to make a few comments on the reality of XenDesktop compared to the position presented by Citrix so far.
First off though I will say that I am impressed by the vision that Citrix presented, the coherence of the vision and the relative openness of the eco system that surrounds it.
I’m not so impressed by the details of their current implementation and the messaging that surrounds it, but I would have been if the messages had been just slightly moderated and developed in the more technical sessions. I’ve made some points about the marketing in a previous post so I won’t repeat them here.
By XenDesktop I really mean the Citrix Desktop Delivery Controller (DDC), it’s integration with the Access Gateway (CAG), the XenServer Virtual Machine Manager (VMM) and the Provisioning Server (PVS).
In summary I like the DDC and the CAG a lot, I like PVS in principle for users who don’t need to install applications and right now I don’t see much to differentiate XenServer.
Some bold claims were made for the XenDesktop, I may have got the wording slightly wrong, but I’m right in principle:
- It’s like a new PC every day
- XenDesktop is applicable to 55% of workers, ie “office workers”
I want to work through these messages and try and get my thoughts straight:
It’s like a new PC every day
The assumption is that a new PC is great and that a new PC every day is even better. The supporting evidence for this is that a PC gets slower over time, because new versions of applications consume more resources and just because of PC entropy. If I get a new PC or just rebuild an existing one the world is rosy. Comments:
- It’s true that PCs do slow down with time, because new apps and web pages consume more resources. But how does XenDesktop help?, most cost models show servers being replaced every 5 years and PCs being replaced every 3 years so that in 3 years time I will have a fast desktop again, but a slow XenDesktop for another two years.
- It’s true that PCs do tend to slow down with time, but speed up following a rebuild, but in my experience that’s because when I rebuild my PC I forgot to put half the applications back on it that I’ve accumulated over the year[s]. I then spend a very frustrating few weeks trying to find and install perhaps the most important 50% of these apps and by the time they are all reinstalled I’m back to the slow PC. So there are two scenarios
- If I have a locked down PC and can’t install anything then PC’s don’t slow down and so “having a new PC every day” makes no difference, because the new PC is exactly like the old one
- If I don’t have a locked down PC and all my changes don’t get persisted from day to day, then I’m going to be really annoyed because one things for sure. I hate getting a new PC because the first thing I have to do to it is to install all my apps and utilities and configure it, and I don’t want to do that every day.
- A new PC every day also assumes I logoff every day, since the only way that I pick up changes to the image with PVS is to reboot this seems reasonable. However this is incredibly unproductive, it’s true that years ago people did close down every application at the end of the day and shutdown their PC, now though its just a quick suspend and resume. At the end of my day I will often have 20 browser tabs open, and 10 applications running, if I have to close all of these down and open them all up again to the same state that will cost me 15 minutes a day off the bottom line, that will wipe out any TCO benefit for sure. In my personal case I restart/logoff my desktop once a month, it runs Vista 64 and its rock solid.
XenDesktop is applicable to 55% of workers
This is a key figure. The logic goes as follows:
- 15% mobile
- 30% XenApp desktops
- 55% XenDesktop users
My view Citrix should have made clear that this is a long term goal, not a current reality. Lets examine the positioning:
- XenDesktop is distinguished from XenApp primarily by the fact that office workers need to personalise their desktop environment. There are six main areas where personalization takes place:
- The roaming profile
- The local profile
- The user hive of the registry (persisted as part of the roaming profile)
- The system hive of the registry
- Program files
- Windows directories
- XenDesktop doesn’t currently deliver a [cost effective] solution for persisting any of these personalisations. Windows roaming profiles and a myriad of third party solutions for profile/environment management do support effectively “roaming profile” persistence
- All of these roaming profile persistence technologies were invented to work with XenApp and are being re-envisioned as solutions for XenDesktop
- So personalisation is not really a reason to use XenDesktop, and since this is the key criteria for positioning XenDesktop and not XenApp as applicable to the 55% of office workers is there another reason? I think there is but its not really specific to Xendesktop:
- Virtual client PCs have higher levels of application compatibility than Windows 2003/8 server – less relevant with Application Virtualization
- Virtual client PCs can leverage the existing management infrastructure being used to manage the physical desktop environment – Citrix dismiss managing Virtual PCs using these legacy management tools though
- Virtual client PCs have a slightly improved fidelity of UI than 2003/8 server – no business case could be built around this difference
- Finally XenApp is cheaper than XenDesktop and so far as I can tell it’s implementation of SpeedScreen and support for multi-media is better than XenDesktop. I feel Citrix need to work harder on getting their positioning straight on this. My feeling is:
- Citrix need to make clear that this positioning is strategic, and will reflect relative investment priorities within Citrix in XenApp vs XenDesktop
- Citrix need to recognise that customers are going to make long term (5 year) investment decisions in servers and licenses and that these will drive them to the cheaper XenApp unless there is a case to do differently
- Citrix need to recognise that for any office worker who does need to install applications or makes changes to the operating system configuration XenDesktop doesn’t provide a cost effective solution and start talking about their roadmap for addressing this
- I also feel that although its not as strategic, it does make sense today for office workers who need VDI and need to personalise their PCs to just boot them virtually, but then manage them as if they were physical PCs. You don’t get the nice conceptual separation of apps, personality and OS, but as I’ve explained above that doesn’t work well anyway with XenDesktop and legacy management is cheap and easy.
Quite a few issues here, but no show stoppers for Citrix, they are on the right path, messaging just needs a little work. of course these are just my quick personal notes, not some details analysis.