Post to Twitter |
Comments (23) |
Views (8615) |
Summary of announcement
Today Citrix announced XenDesktop 4 - a complete desktop virtualization solution to address the needs of all users across an enterprise. XenDesktop 4 includes on-demand apps by XenApp as well as FlexCast delivery technology offering different methods to deliver desktops tailored to meet the performance, security and flexibility requirements of each user. In the future we will also see XenClient to further extend our desktop virtualization strategy to include local VM-based desktops. XenDesktop 4 will be available under a new simpler per user license model more in line to support desktop deployments. XenApp will continue to be available standalone with today's CCU model. We are also providing incentives to move to a XenDesktop license. Since XenApp application virtualization is a critical component of our desktop virtualization strategy we will continue to invest in it with additional capabilities on the horizon as early as the first half of 2010.
Wham! There you have it, finally a single product option. Something that I have wanted for a long time to address a far broader range of use cases and truly enable desktop virtualization that goes beyond a simplistic one dimensional VDI view of the world.
So what are those use cases and drivers?
I've written in a past blog the time is now! about why I feel WIndows 7 is a long awaited stimulus to drive the next desktop refresh. In addition, the analyst data I have seen strongly indicates that 2010 will be the year of the PC refresh after years of skipped upgrades. This I believe will lead to many more people seeking to understand how desktop virtualization fits into to their strategic plans to ensure that they make the right investments in 2010 for their IT infrastructure as the economy begins to recover.
When I talk about drivers for desktop and application virtualization to customers from CIO to hardcore techie, I generally find that everything fits into three buckets. Here are some common perspectives that I have shared that resonate with our real world customers who are implementing at scale today.
Business Perspective
- Simplify business continuity and build it into the core architecture.
- Enable quicker and easier office moves and enable mobility from any connection from a broad device set.
- Invest in M & A and global expansion.
- Enforce stronger data standards and security through centralization.
- Drive increased consistency, more efficient staff based on location.
- Enable outsourcing.
User Experience Perspective
- Ensure consistency of user experience across any network. WAN and bandwidth matter.
- Improves performance when latencies are managed down by moving apps closer to the desktop.
- Enable telecommuting and access from any device/connection.
- Recover faster from faults to increase productivity.
- Introduce new productivity models like BYOPC.
Technology Perspective
- Leverage power and cooling efficiencies in the data center to reduce costs.
- Consolidate data centers and extend reach of existing data centers.
- Build greener user buildings in metropolitan areas and reduce carbon foot print.
- Reduce complexity in workplace and datacenter management OpEx by reducing the # of instances to manage.
- Ensure reliability of simple clients to reduce helpdesk calls and end user break fix visits.
- Reduce management costs and risk for infrequently connected devices such as laptops.
Once people start to understand the potential from their respective points of view, it usually boils down to desktop virtualization is a way to drive costs down over time and increase productivity very quickly.
To solve for the above use cases with the current distributed computing model is very difficult. Primarily this is because at scale distributed computing is complex to manage with a lot of overhead and many moving parts. This then leads to many points of control to make a change - flexibility is next to impossible. All this adds up to slow time to value for anything that needs to get done quickly, and hence the model is not very agile.
When it comes down to picking the right technology option to enable a new desktop model, the predictability and performance over a diverse network infrastructure become key considerations. Many customers realize that to reduce support complexity and achieve service delivery consistency it is better to use a technology that addresses the majority of your use cases. If you truly understand the above use cases, it becomes obvious that to truly leverage your investment, WAN and smart utilization of bandwidth is a must. I'd argue it is the lowest common denominator, and it takes more than a protocol to deliver the best possible user experience. For example, working in the office you may be on a LAN or MAN depending on the location of your data center, but when you travel or go home network latency and bandwidth matter. If you have to switch display protocols with varying bandwidth requirements from site to site then the complexity of supporting this when a user calls just erodes the cost benefits. In addition, some customers simply reject the idea of implementing solutions that lock them into proprietary client hardware solutions, or hypervisors as they lose price leverage. Many customers have told me that they want technologies that let them choose what type of client device to run on and hypervisor choice. That could be a Thinclient or simply a repurposed commodity PC, but with a mature protocol that is proven to run over diverse networks and uses bandwidth intelligently. Similarly these customers understand that hypervisor diversity is inevitable and so want to invest in management layers that support this coming trend.
Today Citrix has a range of HDX technologies to address a plethora of user experience use cases. We also run our solutions on multiple hypervisors and on physical hardware. This is why we are winning large XenDesktop customers including a 100,000 seat deal that we recently closed.
Is XenDesktop needed since XenApp also enables desktop and application delivery?
Not having XenDesktop 4 resulted in an artificial TS vs. VDI debate thanks to Citrix 
that has just continued to brew. Yes it's true, XenApp can host desktops and apps on a server operating system, so this leads to the logical question why do I need XenDesktop? In a blog last week on Brian Madden's site this sentiment was highlighted once again. I believe this is now a moot debate. With XenDesktop 4, it really does not matter which model you choose. What's more important to understand is that you choose the right model to address your business need that fits the right economics and time to market for you.
Despite my pre-XenDesktop implementation experience and choices due to technology availability, I've always wanted and believed in a move towards a desktop OS. Why? It was largely driven by:
- It's a desktop. It makes the most sense to deliver it with a desktop OS and avoid any issues or optimizations that may come up in the future that I may not have been aware of.
- App compatibility is not a problem on a single session desktop OS. Yes you can use 1-1 XenApp, but for reason 1 I still prefer a desktop OS.
- 3rd party vendor support is not an issue on a desktop operating system.
- Consistent service delivery of running a desktop OS across all use cases. This includes users being familiar with their Laptop OS - corporate or personal.
- Reason 3 above makes it easier for in house developers to adopt.
Now that said, I've lived through large scale XenApp desktop and application deployments and seen that the app compatibility issue is marginal after 20 years of Citrix pioneering the Server Based Computing model. Most vendors support their software on multi user operating systems like Window 2003 and 2008 and XenApp 5 Feature Pack 2 now addresses those edge cases by enabling VM hosted applications delivered from a desktop OS.
However there is a twist. In my opinion this is the advent of Windows 2008 R2 - only available as a x64 operating system. This presents several additional considerations.
First, how fast is the world going to move all their desktop applications to a x64 operating system and have the hardware on the backend to host x64 desktops and apps? Note I did not say application compatibility, because based on my experience most x32 applications run just fine on a x64 Windows operating system. There is some repackaging and testing to be done, and there are of course exceptions like apps with 16 bit installers etc. but in general I have not seen compelling evidence to suggest that this will not work for the vast majority of apps.
Second, Windows 2008 R2 and Windows 7 share the same code base. At the technical level their kernels are the same version. Thank you Microsoft for finally pulling this off! This will vastly improve application compatibility. At a very high level of assurance I can be confident that applications re-packaged for Windows 7 will run just fine on Windows 2008 R2. Now customers can choose what is most comfortable based on their requirements (well they will once we release XenApp for x64 which is slated for beta towards the end of 2009) because it really does not matter - their investment moving to either operating system is well protected from an application compatibility perspective.
Finally, for users on XP or Vista, I don't know of anybody new who at this point will adopt desktop virtualization by migrating to Windows 2003 given the effort to get all your applications re-certified and then do it again when Windows 2003 will reach end of life in a few years. It's just not worth it. Windows 2008 x32 is certainly an option, but again to ease the application migration effort it seems more prudent to me to get your applications ready for Windows 7 and Windows 2008 R2 to future proof yourself and have application portability between the operating systems.
Now I fully expect many people reading this to say, but I want to do X with operating system Y for reason Z. That's ok, and clearly as I elude to above, given the world is so diverse it's foolish to assume a single prescription and proclaim this is how you do it. As I illustrate below, XenDesktop 4 let's you choose the best model for solving real world business problems. I like to think of it as a sliding scale that is a pragmatic realization by Citrix that customers want many ways to skin a cat. I believe this flexibility will enable our products to be more easily consumed now that every use case can be addressed irrespective of your OS choice, application compatibility concerns or x32 vs x64 belief system - all with the most predictable user experience powered by HDX.
Key considerations and takeaways
Virtualization is already forcing a datacenter re-architecture. I would argue that those who do not believe virtualization is a force driving consolidation in the data center that has past the hype and inertia stage are in the minority. As this shift happens it is very important to design your virtual infrastructure and organization to handle desktop scale and service levels. I've blogged about this in the past desktop virtualization is not server virtualization. And I can't begin to emphasize enough how important this point is. It's a mistake to think desktop virtualization is a simple extension of your existing server virtual infrastructure.
The forces of globalization, offshoring, teleworking, mobility, and green are causing more users to be mobile. Forces such as consolidation, data security, business continuity, and green are driving us towards centralization. Business are becoming more complex and diverse, and the distributed computing model will only get more expensive to manage and is not designed to handle the needs of an agile organization that requires a lot of flexibility.
As a result, XenDesktop 4 is a landmark release in our history. It brings together the best technologies and reaffirms our commitment to enable customers to deliver IT as a service with desktop and application virtualization. XenDesktop 4 demonstrates how at Citrix we understand that desktop virtualization is so much more than just VDI. This is at the very heart of where we have come from as an organization over the last 20 years. It's in our DNA how to deliver user experience over diverse infrastructure, and this is just the beginning. We continue to innovate and expand our reach through diverse devices, HDX, new delivery models such as Dazzle, continued investment in application delivery and so much more to come. It's an incredibly exciting time for us at Citrix, now that we've stepped up to heal the broken hearts of TS vs. VDI and enable a new tomorrow that represents pragmatic choice.
Comments (23)
Oct 06
Robert Morris says:
While I think most will disagree with you on some of your OS choice comments, an...While I think most will disagree with you on some of your OS choice comments, and reasons why some clients will still be deploying 32-bit Win2003, the bottom line is flexibility.
This new initiative is exactly what we architects in the field need - allowing us to provide the best solution in place, for any given situation. It does not force us to choose VDI over TS, or worse yet - design two separate infrastructures leading to even more support overhead. Now we can simply design the best application delivery infrastructure for any given client needs, without regards for which product to choose. The title of your article should have been "less rules...more fun".
This announcement would have been a homerun had Citrix chosen to stick with a concurrent user model. The decision to use a per-user model is the only thing keeping this from being the ground breaking announcment Citrix would desire. Time will tell if this was the right choice for Citrix, their partners, and ultimately the customers.
Overall, a great announcement, and your post was worthwhile sharing of both your vision and past experiences. Thanks for taking the time to reach out to the community with your thoughts.
-Robert Morris
rmorris@agsi.us
Oct 06
Andrew Wood says:
but why have a concurrent model for XenDesktop - all of the features available i...but why have a concurrent model for XenDesktop - all of the features available in XenApp are available in Xendesktop as of 4.0
Likewise all the "features" in Xendesktop are available in XenApp as of FR2 with XenApp 5.0.
The difference (as I see it) is in the licensing model - XenApp is concurrent; XenDEsktop is per user. So the 'premium' of license cost for XenApp pays for the 'flexibility' of concurrent use. Potentially you can name users (say for desktops) in which case you can apply xendesktop licenses
Perhaps (cleverly?) - it allows for yet another Citrix marketing exercise for a product name change - XenConcurrent and XenNamed?
Oct 06
Anonymous says:
'Per Named User' is indeed keeping this from being the big event that it should ...'Per Named User' is indeed keeping this from being the big event that it should 'have' been. Customer environments are diverse and for many, i would speculate, including myself, this per named user approach has just raised the aquistion cost from what would have been 800k to 'several' million.
Oct 06
Anonymous says:
Sorry, but VM-Hosted apps is not ready for prime time. Until then, y...Sorry, but VM-Hosted apps is not ready for prime time. Until then, you've taken away a valid use case for XenDesktop and replaced it with something subpar and unusable.
Oct 06
Mike Stanley says:
Yep, unless Citrix will continue to offer XD at a CCU license for educational ma...Yep, unless Citrix will continue to offer XD at a CCU license for educational markets, this per user license change is a deal-killer for us.
Our cost to implement OS Streaming to our 1,000 computer lab machines just went from a potential $195,000 to $6,300,000.
Our state, like most others, is broke. A 32x price increase means we can't buy your product.
We were looking at showcasing XenDesktop in our labs to prove it could work everywhere else - faculty and staff desktops, remote access, you name it.
Oct 06
Andrew Wood says:
Anon - but VMHosted apps *is* effectively XenDesktop - as far as I under...Anon - but VMHosted apps *is* effectively XenDesktop - as far as I understand it - effectively
I'm not quite sure how Mike's done the math - but I agree moving from XD on a CCU basis to (say) XenApp using VMHosted is going to be more expensive - but not x32 more
However, "more" is in indeed "not less"; a VECD license is per device isn't it - that would have been a better model and kept Mike's costs in the same ballpark
Oct 06
Mike Stanley says:
I'm talking about the cost increase for XenDesktop from what it would have been ...I'm talking about the cost increase for XenDesktop from what it would have been last week to what it is today.
With 1,000 lab computers but 28,000 unique users (a little more than that, actually), XD3 would have cost, for the Advanced Edition, 195 x 1000 = $195,000.
With those same 1,000 lab computers and only doing OS Streaming, so only ever (for now anyway) 1,000 possible concurrent users, but paying for XD4 licenses on a per user basis, the Enterprise Edition (minimum required to do OS Streaming) would cost 225 x 28,0000 = $6,300,000.
6,300,000 divided by 195,000 is 32.x That's how I got 32x more.
I'm not talking about moving to XenApp. I have physical desktops I want to stream an OS to, and this per user licensing change literally destroyed XenDesktop as an option for me. Could I possibly pay twice as much for XenApp and get similar functionality from an app virtualization standpoint? Sure I could - but I'd still be left with 1,000 physical desktops to manage.
Oct 06
Anonymous says:
Forgive me if I have this wrong, but XD4 just seems to be a relaunch of a few ex...Forgive me if I have this wrong, but XD4 just seems to be a relaunch of a few exisiting Citrix products all bundled together. It looks like XD3 with the flash enhancment, XenApp5 FP2, and some marketing rubbish i.e. Flexcast. All of the above were available with XD3. Maybe I'm missing something?
As for changing the licensing model.......well that is just crazy!! Desktop virtualisation via VDI is already struggling to take off due to high capex costs, and this will just be another reason to dismiss it.
And I won't even mention the number of customers already thinking of giving up their subs
It seems Citrix have once again rushed out a load of marketing catch phrases without any real substance behind them.
Oct 06
Anonymous says:
Please correct me, if I am wrong: If I choose to go with a VDI approach, e.g. X...Please correct me, if I am wrong:
If I choose to go with a VDI approach, e.g. XEN Desktop, I have to buy Microsoft VECD licenses for every potential client connecting to my Virtual Desktop server farm - every year.
In contrast, if I choose to virtualize applications instead, e.g. via XenApp, I do NOT have to buy Microsoft VECD licenses at all. The difference can be significant over time.
Again, please let me know, if this is correct, and why I would want to use Virtual Desktops instead of Virtual Apps, when it comes to cost.
-Alex
Oct 06
Anonymous says:
Alex, Correct. So maybe When you add XD and VECD together. $350 list plus...Alex,
Correct. So maybe When you add XD and VECD together. $350 list plus $120 at the high end for VECD=$470 Is $130/per CCU more for XA PLAT worth the convenience of CCU vs. named user? I think so if I have 5000 users but only need 2500 to connect at a time.
Oct 06
Anonymous says:
Thanks for your response. With this pricing, the term "Platinum Partner" g...Thanks for your response. With this pricing, the term "Platinum Partner" gets a whole new meaning (-;
If I have, say 250 CCU out of a total of 7000 users, and my CFO wants to save money, and I have to present apps - not desktops -, then XA seems the far more reasonable choice to me. Any thoughts on this?
Thanks, Alex
Oct 06
Bryan Mann says:
I think this may drive more users back to XenApp and the central idea that the A...I think this may drive more users back to XenApp and the central idea that the APP is what matters not the desktop/device. Consider $350 list for XD PLAT plus VECD PER YEAR per named user. How close does that get me to the $600 for XenApp PLAT and the ability to buy per CCU instead plus no yearly cost. OR drive them to View.
Oct 06
Anonymous says:
Applications make people productive, not the Operating System, nor a Desktop, or...Applications make people productive, not the Operating System, nor a Desktop, or desktop environment the apps are presented on. A desktop is the interface between the Operating System, and the productivity tool (the application). In a virtualized world, is the high licensing cost portion of this abstraction layer warranted? The amount of time spent working with applications is far greater per average user per day, than the time spent navigating/customizing/etc. the desktop. If I compare the total cost of a Virtual Desktop vs. a Virtual App solution based on cost over productivity, the Virtual App scenario appears way more economical, because I don't have to spend perpetual licensing money on something that is not a core productivity tool...
Please tell me, if I am missing something here, because I don't see the merit of the VDI cost model, mostly due to the recurring VECD licensing cost, compared with an RDP/TS, or Virtual Application model, when all is said and done.
Thanks, Alex
Oct 06
Anonymous says:
If it's a desktop use case, how is this any different than desktop products in g...If it's a desktop use case, how is this any different than desktop products in general? It seems that XA can address most use cases, so I agree this will drive people towards that if CCU is important to them.
Oct 07
Garrick Stafford says:
In my opinion, I do not think Citrix has finalized its pricing strategy for the ...In my opinion, I do not think Citrix has finalized its pricing strategy for the new XenDesktop 4 product line. A Citrix representative contacted me yesterday and indicated that Citrix will be offering a 2 for 1 deal with respect to the XenDesktop licensing. For each XenApp license that will give 2 XenDesktop licenses.
Based on the above comments regarding pricing, my hope is that Citrix will offer more attractive pricing options, even better than the 2 for 1 deal. Please keep in mind that this information was given to me yesterday but is subject to change at any time until Citrix officially releases the true pricing for XenDesktop 4.
It seems that Citrix is being more aggressive at trying to make Citrix a total application, server and desktop delivery solution. This was more evident when I asked the Citrix rep if there was any improvements to their Profile Management product that would allow users profiles to roam from non-Citrix desktops (XP, Vista, Win7, etc), XenDesktops, Citrix XenApp and Streaming apps (offline and online) and be able to handle system drive letter differences in either case. He then mentioned that one solution would be to move everything to Citrix!
Oct 09
Anonymous says:
"indicated that Citrix will be offering a 2 for 1 deal with respect to the..."indicated that Citrix will be offering a 2 for 1 deal with respect to the XenDesktop licensing."
But dont forget that 2 for 1 deal comes with with a doubled Subscription Advantage fee.
Oct 11
Anonymous says:
Actually the SA does not double. Folks need to slow down and look at the number...Actually the SA does not double. Folks need to slow down and look at the numbers. Look at the numbers.
Oct 13
Anonymous says:
Bottom line, Citrix needs to allow for a CCU model until people are fully VDI an...Bottom line, Citrix needs to allow for a CCU model until people are fully VDI and then they can convert to CCU or just buy more CCU. We need a path to get there, why are you folks killing our ROI case. We fight for you everyday and you turn around and pull this model on us. It does not help, so PLEASE do something to help us help you....
Oct 13
Anonymous says:
Why is not Citrix offering a single product Xen(BlaH) that can deliver apps and ...Why is not Citrix offering a single product Xen(BlaH) that can deliver apps and desktops from both TS and VM based platforms in both named user and CCU based licensig modesl? why do they still have XenDesktop and XenApp? One product with multiple licensing models will allow partners and resellers to focus on selling one product and will be easier for customers to understand.
Oct 19
Anonymous says:
As another .edu user, here are the thoughts I sent to my sales rep: v\:* Unknow...As another .edu user, here are the thoughts I sent to my sales rep:
v\:*
Putting it bluntly, the new licensing scheme is a deal killer for us rolling out desktops to students and our labs. There's no WAY at the current license price we will be able to afford to roll out 3000 licenses, and no sense for us to even continue looking at the Xen Desktop product line: we simply can't afford it. Our labs are rarely more than 40% full (if that) and we never have 100% of our students using labs at ALL during a semester much less concurrently. The 'license every user' model might work in the enterprise, but for the education market it is utter nonsense and is a massive price hike.
In essence this price increase (and no matter how Citrix spins, it IS a price increase) will likely drive us to look at other VDI products. The pricing is simply untenable.
Oct 20
Harry Labana says:
Thanks all for the passionate feedback. As a result per device, CCU and educatio...Thanks all for the passionate feedback. As a result per device, CCU and education options have been added. See http://community.citrix.com/display/ocb/2009/10/20/Thank+you+-+XenDesktop+4+makes+desktop+virtualization+real+for+a+broader+set+of+use+cases+now%21
Oct 20
Anonymous says:
This is more like it. Now you can really claim to be industry leading. Thanks fo...This is more like it. Now you can really claim to be industry leading. Thanks for listening, it does matter even if a small % will still not be happy. Looking forward to innovation to take this to the next level.
Nov 01
Anonymous says:
Well with the new Citrix Xendektop licensing model, they have now added per devi...Well with the new Citrix Xendektop licensing model, they have now added per device licensing, so for those edu users etc, this would indeed be a better use case? Would it not? Per device licencing will still allow those Organizations to use say 1000 devices amongst 3000 students.
Add Comment