Simon Bramfitt

Desktop Virtualization Performance Testing

Citrix released their latest XenDesktop single server scalability test results today; showing a dual socket Intel Xeon x5570 (Nehalem) server with 72GB RAM running 130 Windows XP VMs per host. This is a significant improvement on Citrix’s previously published results, and certainly enough to bring it back into line with results VMware previously published. 

More important than the results though is the test methodology that Citrix has adopted this time around.

 For the past year I have been trying to encourage desktop virtualization vendors to adopt a standard testing methodology for virtual desktop workloads and I am very happy to be able to say that it looks like Citrix has come into line with my way of thinking. Rather than relying on an in-house developed test, Citrix have adopted Login VSI 2.0 from Log•in Consultants as the core of their testing process. Login VSI is a specifically designed benchmark for server-based computing and desktop virtualization environments. You may know it better as part of Project Virtual Reality Check the joint project by Log•in Consultants and PQR. Project VRC is the only fully independent benchmark of desktop virtualization workloads running on diverse physical and virtual platforms. By establishing themselves as trustworthy independent subject matter experts, the Project VRC team has benefited not just their customers but a global audience of IT organizations who can take advantage of the research performed under the Project VRC banner. Even more importantly the breadth and quality of information presented through Project VRC is acting to improve the quality and credibility of all server hosted desktop virtualization vendor issued performance benchmarks.

 Now that Citrix have seen the light, I hope that all server hosted desktop virtualization vendors will follow suit and adopt Login VSI as their testing benchmark. Anyone who doesn’t should not be surprised if their performance data is called into question.

Straining upon the start. The game's afoot:

"                                    And you, good yeoman,

Whose limbs were made in England, show us here

The mettle of your pasture; let us swear 

That you are worth your breeding; which I doubt not; 

For there is none of you so mean and base, 

That hath not noble lustre in your eyes. 

I see you stand like greyhounds in the slips, 

Straining upon the start. The game's afoot: 

Follow your spirit, and upon this charge 

Cry 'God for Harry, England, and Saint George!'"

Shakespeare. Henry V: Act 3 Scene 1

Today desktop virtulization comes of age with the release of VMware View 4. Now at last we will see side by side both Citrix and VMware's visions for the future of desktop virtualization; and while not quite as historic an event as the battle of Agincourt, it is clear that the lines have been drawn and that the future of these two companies will be shaped based on how they fare in the battle to own the virtual desktop.
Citrix and VMware have very different perspectives on how to deliver against corporate IT needs (and to a significant degree on what those needs comprise of), not to mention each having a complex network of alliances that would make many aspects of European history appear transparent. So the coming year will be an interesting mix of each companies own capabilities as well as just how close the alliances they have forged have become. I'm looking forwards to an interesting year.

Now that we are starting to see the battle lines being draw, some small part of me is secretly hoping that Citrix's own Harry is rousing his troops with a cry of 

"Once more unto the breach, dear friends, once more;
Or close the wall up with our Citrite dead."

Licensing Storm in a tea cup

Social media presents corporations with as many opportunities as it does challenges. You need look no further than the recent criticism that Citrix have received regarding the announcement of the XenDesktop 4 licensing model to see just what happens when social media isn’t effectively managed. 

Given the level of dissatisfaction expressed over the past week, a bunch of Citrix's execs will no doubt have been sitting behind locked doors this weekend trying to work out a way to mollify their most vocal complainants without leaving themselves open to even more criticism. I fully expect that within 48 hrs we will see some sort of announcement from Citrix to the effect that a new per device licensing scheme will be offered along side the ‘named user’ model previously announced.

The named user model is the most logical solution to licensing a comprehensive VDI platform. I’ve been through this before, as has Chris Wolf and rather obviously as did a significant number of Citrix’s leadership team over the past year or so. But this will not stop the announcement of a change to XenDesktop 4 licensing in favor of a model that will in the long term create more problems than it will solve. 

Citrix response to XenDesktop 4 Licensing Concerns

I received an interesting response to my previous post on XenDesktop 4 licensing from John Carey (Citrix Director of Licensing and Maintenance Programs).  As not everyone reads comments I'm reproducing it below in full (all spelling errors can be put down to John writing this very late at night on an iPhone). 

Simon, as always interesting and insightful comments. I'd like to clarify some potential misconceptions that have gained credence in the blogsphere. With the introduction of XenDesktop 4 and the User license model the XenDesktop product and subscription advantage pricing were actually revised downward. At the same time the existing limitations on application virtualization within XenDesktop were removed significantly increasing the value of the product to customers. The unshackling of XenApp in XenDesktop 4 IMO reframes the pricing conversation away from XenDesktop ccu and puts it in the context of XenApp ccu. XenDesktop 4 is highly discounted whn compared with XenApp (up to 50% lower SRP) We should also note that a single user of Xendesktop 4 is licensed to consume an unlimited number of virtual desktops and an unlimited number of virtual applications from an unlimited number of devices, online and offline. Citrix has also publicly stated our ongoing committment to invest in and innovate the XenApp product line sold exclusively under a concurrent user model. In fact for high concurrency use cases [ I think John means 'low concurrency' here] and remote access scenarios, Citrix continues to recommend XenApp sold under a concurrent user license model with its' published desktop feature, appliaction isolation and streaming features and (with the release of XenApp 5.0 Feature Pack 2) VM hosted applications feature. As for trade-up yes in a 2:1 scenario the SA renewal cost increases. However the per unit cost for SA is lower than XenApp, the increase is due to the doubling of licensed users available to XenApp customers at up to 80% savings on XenDesktop 4 SRP. Obvioulsy I'm invested in clarifying our position, I'm also conscious that customers have very real concerns around immediate cost and budget impacts of extending their virtual desktop deployment, Citrix continues to be invested in helping customers solve these and other challenges with the most cost effective and hi performance solution available in the market today.

John Carey

So to summarize.

• Citrix reduced the price of XenDesktop licenses and Subscription Advantage as part of the transition to user based licensing, so that a XenDesktop 4 license is up to 50% less expensive than a comparable XenApp license
• Citrix are continuing to look to ways to help customers with concerns about the cost of transitioning to the user based licensing model
• XenApp will continue to use the concurrent user licensing model and Citrix is committed to continuing to develop the platform.

I confess that John's first point XenDesktop 4 licensing and SA price reduction has passed me by (I'm more driven by value than price). When you consider that a XenDesktop 4 can behalf the price of a XenApp 4 license,yet XD4 includes a XA license it becomes clear that those Citrix customers who have heavily invested in XenApp in the past and run at high concurrency might see a reduction in their Subscription Advantage costs.  And it's clear from the second point that the story is not yet over, Citrix is continuing to look at ways to help retain customers like Mike Stanley.

A small but valuable incentive - Updated

There is one small but valuable incentive to migrating to XenDesktop that might make existing XenApp administrators happier. Those organisations that have grown organically over the years and so finished up with many separate licenses, will finally have the opportunity to consolidate them into a single new license - that along might convince some people I know that an upgrade to XenDesktop 4 is called for.

Take a look at this small section of the overall list of licenses that need to be managed for a large environment that I support. Replacing all those blocks of 50 licenses with a couple of new 1,000 license blocks would be heaven.

 One point to consider should you take this route would be to avoid the temptation to place all your licenses in a single block. If at some point down the road you want to stop Subscription Advantage on half of your licenses because you no longer need them, it will be a lot simple to do so if you have a single block of licenses that you can point to.

_______________________________________

This post was updated to correct an error on my part where wrote  'License Files' where I meant 'Licenses', and extended to provide a fuller explanation of the benefit this offers. 

More on VDI licensing

In my previous post exploring licensing considerations for a VDI licensing model. I came to the conclusion that the concurrent usage licensing model had significant shortcomings compared to a user based model. However, there are other factors that need to be taken into consideration. While a private poll of 500 of Citrix's customers showed broad support for the XenDesktop 4 licensing model, Mike Stanley's previous comments highlight the two most important secondary factors that should have to be taken into consideration when deciding on the upgrade path.

The most visible of these issues is cost. Transition to user based licensing will inevitably result in an increase in the number of licenses required. While many (perhaps most) existing customers will be able to trade up to XenDesktop 4 for little immediate cost, the on-going Subscription Advantage commitment will be substantially more than their previous outlay. Introduction of a significant increase in licensing costs (even if it does not apply for another 12 months) is going to raise serious concerns and cause many customers to review their commitment to Citrix. I think that Citrix need to look closely at how this increase in Subscription Advantage fees will impact their customers and consider taking action to limit its impact. The global recession is still going to be depressing IT budgets for some time to come, and far sighted IT managers will be looking closely at what an increase in the number of licenses under Subscription Advantage will cost. Using a 2 for 1 upgrade path as a way to boost Subscription Advantage revenue is not going to be well received. A reduction in the unit cost of Subscription Advantage would be a welcome announcement at this time, not to mention prompt action to address the concerns of customers like Mike where a 2:1 conversion ratio is of very little benefit.

Which brings me to the second point: Trust. Mike asked a very important question in his previous comments.

Why buy into an expensive XenApp solution today, knowing that the next version may switch to per user licensing ...

Mike Stanley

Now that the door to user based licensing has been opened, Mike is just one of many asking what next for XenApp.  Personally I don't see Citrix making any changes here.  They had the opportunity once when off-line Application Streaming was introduced but didn't take it, acting to do so now is most unlikely. Having said that, I do think that Citrix have a problem here. Just as Citrix and Microsoft are seeing greater adoption of XenServer and Hyper-V as VMware's customers becoming disaffected with their licensing practices, Citrix need to ensure that they do not loose market share to competition in the application and desktop delivery space. A clear message to reassure customers that the XenApp licensing model is not going to change would be most welcome.

How do you choose a VDI licensing model? - Updated

Citrix's announcement that XenDesktop 4 licensing will break from the concurrent user model in favor of licensing by named user has got many people up in arms,and making loud claims as to how this will force them to abandon plans to adopt VDI. However once you get past the shock of the new and start to look at things more coolly it soon becomes clear that while it might have be possible for Citrix to retain a concurrent licensing model a little while longer, the direction that the VDI market is heading in will force every vendor to move away from the concurrent licensing model in the near future. 

I was going to write about VDI licensing from the customer perspective, but the Burton Group's Chris Wolf beat me to the punch on his blog (see side bar for link) 

Reworking licensing to be user-centric needs to be a top priority among client application vendors heading into 2010. Vendors that insist on binding licenses to physical devices in an increasingly virtualized world are not part of the solution. They’re part of the problem.

Chris Wolf

so I'm going to take a look at this from the vendor perspective.

Even today XenApp 5.0 no longer operates with an exclusively concurrent licensing model. The introduction of support for off-line streaming apps requires a hybrid licensing model where hosted apps are served from a pool of licenses assigned concurrently, but streamed apps use license leasing. This is necessary because streamed apps can be configured to continue to operate when disconnected from the host network, and at that point it is no longer possible for the License Server to determine if an application is being accessed used or not. The only viable solution to this problem is to adopt a leasing model. Allocating licenses to the user for a fixed duration with the option to renew for as long as the streamed application is being used. This by the way does more than ensure that Citrix maximizes its licensing revenue.  By preventing a license from be reallocated until its lease expires it prevents licenses from being double booked and so ensures that a customer can meet any audit with confidence.  I can think of a few companies than have been subject to audit and found themselves on the wrong side of a multi-million dollar settlement who might wish that other vendors would offer that approach.

Looking forward beyond the launch of XenDesktop 4 to the day when XenClient is launched, the technical landscape becomes much more complex.  One user might be running a VDI desktop hosted within the data center, another might be a task worker running a single hosted application. Others might be working from their homes running a VDI guest on a XenClient on an otherwise unmanaged workstation but with a VPN connection to the corporate network, while someone else is running another VDI guest on a laptop also running XenClient fully disconnected from any internet connection.  You can no longer control licensing based on connection count if your product does not require a connection to work. This change applies not just to Citrix, but to every VDI software vendor that intends to compete across the full gamut of VDI technology platforms. 

If concurrent licensing is no longer an option that leaves per machine or per user. In most environments, there isn't much difference between these two models. In most cases the number of people sharing PC is very low, so a PC based model could work just as well as a per user model, and this is where it gets interesting. If you go back to Citrix and ask why they adopted the concurrent licensing model, you might if you were lucky find someone who has been around long enough to be able to remember the initial reasoning behind that decision which if I remember correctly went something like 'We license our technology based on usage. You are only using ICA when you have an active connection between PC and server, so we think it is only right that you only consume a license when you have an active connection to the server.' Now the intellectual property that you use is far more than just the presentation protocol and much of that intellectual property is still being used even when the connection is not. But the focus is still on the user of the technology rather than on the machine that it is consumed from.  Furthermore, with the continual increase in diversity of access platform many of which might be used in a single working day, leasing to a device would lead to multiple licenses being consumed by a single user in the course of a day and/or licenses being orphaned as a user accesses stateless devices that are re-provisioned from a gold image every time they are used. So by default licensing based on who is using able to use XenDesktop is the only choice left open.

Unfortunately there is never a good time to move to a new licensing model, Citrix could have forced through the change with XenApp a couple of years ago with the introduction of off-line access to streamed applications, but rightly, in my opinion, chose not to. Leaving the change until VDI is more mature would create even more anguish than it has done today.  So now, as the VDI paradigm starts to gather speed, is probably the least bad time to make this change. 

What matters most at this point is that the transition to this new licensing model be handled with appropriate care and respect for the current customer base. Citrix's opening offer is very attractive to existing customers once you look at the detail and compare upgrade costs with the cost of Subscription Advantage to maintain a XenApp subscription, especially when you take into account that if you time the upgrade right you will finish up with both the XenDesktop 4 upgrade and the XenApp 6 (Parra) release within the same 12 month period. The 2 for 1 offer works very well for most existing XenApp customers but not all. Citrix need to look at how it can address market segments such as retail, education and healthcare where the user to device ratio is frequently much higher.

My final concern is that Citrix ensure that an appropriate balance is achieved between the cost of Subscription Advantage going forwards and the value delivered when customers find themselves owning a far larger number of licenses than they might previously have had.

XenDesktop 4 Licensing

Citrix's decision to change to name user licensing for Xendesktop 4 is understandably creating a bit of a stir (see the comments here and here). After having lived with a concurrent licensing model for so long it is easy to forget that a named user model is for most organizations a better fit than a concurrent user model, and is the model used by both MS and VMware. Nevertheless the transition to another licensing model is bound to be a little difficult for both Citrix and their customers, and inevitably some will win and others will loose.  

Citrix have provided a cushion to ease the transition for most users by offering a 2 for 1 migration package where every XenApp 5 license can be traded in for two XenDesktop 4 licenses. This should address most customers needs, but for some it clearly isn't going to be enough.  Anyone with a large differential between concurrent users and total users is going to complain, and some will probably consider jumping ship to other technology vendors.  However, this is not the end of the story.

In all the excitement of this announcement we should not forget that XenApp is not being discontinued and there is no requirement to migrate to XenDesktop 4. XenApp is still licensed by concurrent user and can still be used to deliver a Published desktop just as it could last week. And while I think that XenDesktop is becoming Citrix flagship product.  Let's get this absolutely clear, XenApp is not going away.  The successor to XenApp 5 will be released as a tech preview some time in Q4 this year and barring any possible name changes we should expect to see XenApp 6 released in Q1 next year. Citrix don't share everything with me, but I fully expect to see dedicated application delivery products being developed by Citrix for a very long time to come (25 years should just about cover it).

For those customers who still find the cost of implementing XenDesktop4 unpalatable, the best advice I could offer would be to wait and see. Citrix's Sumit Dhawan has already made it clear that what was announced on Monday is not the final picture, and that changes will be announced to address customer concerns. 

"Finally, if you are in an industry that has high concurrency ratios - stay tuned - we are working on offering you the right options to address your requirements."

I won't speculate as to what these changes might be, but to all those people who are thinking that the change to user based licensing is unacceptable, I'd offer the following advice:

Think of the prices that Citrix have announced are bargaining positions only.  If you don't like them you can walk away or you can negotiate a better deal. Given the current economic client, Citrix's desire to get ahead of VMware in the number of VDI seats sold, and the fantastic opportunity that XenDesktop 4 represents; I'd suggest that walking away is probably not the right move.  

I wonder, should I write up a post on how to negotiate a VDI license agreement?

XenApp Deathwatch (updated)

A little while back I suggested that Citrix should stop selling products and start selling services, specifically an app delivery service that and a desktop delivery service combining the best elements of XenApp, XenDesktop, Provisioning server for Desktops as a single suite that could be used any way that the customer wanted to deliver either a desktop or an app, hosted or streamed whenever and however the customer wanted.

Well today it seems that Citrix have gone one better. Not content with giving XenApp admins access to XenDesktop to host desktop OS dependant applications, Citrix announced today that the VDI guys can now have piece of the application delivery action by integrating XenApp licenses into what is fast becoming Citrix’s new flagship product XenDesktop 4. Now with XenDesktop 4, one single product license allows you to stream or host a desktop or an application anyway you want. I just hope Citrix don’t forget where the idea first came from - I’ll be expecting a check in the post Mark. 

Aside from opening the door to total delivery flexibility, the other really big change is that for the first time ever Citrix has stepped back from the concurrent usage licensing model.  XenDesktop 4 licensing follows the Microsoft model of one license per user.  The reasoning behind this is not clear and is likely to create some angst amongst Citrix diehards, but old time WinFrame admins are well versed in adapting to change. It will be especially difficult for the channel to explain why XenDesktop 4 is priced per named user while XenApp and XenDesktop 3 are both still priced per concurrent user. 

There's still one piece of the picture missing today, but with a clear indication that XenDesktop 4 will integrate with XenClient just as soon as it is launched, Citrix will have a complete application and desktop delivery suite available mid 2010 just in time for its customers to have got to grips with the idea that desktop virtualization is about to play a big part in their future.

I can’t say that I’m 100% happy though; XenDesktop 4 is still only licensed in terms of pre-paid licenses. There’s no service based offering providing capacity on demand or monthly billing, so you still have to get your demand forecasting right and endure that you don’t over or worse under estimate the number of licenses you need. But there's plenty of time for that in the future.

Here's Tom Flink explaining the announcement to the sales channel, makes me want to run out and buy it right now :)

So is this change the beginning of the end for XenApp? I’m sure we will see XenApp 6 announced towards the end of this year, bringing with it a raft of new management features, (Update  - I have solid assurances from Citrix that XenApp is not going to go away and that XenApp will continue to be enhanced well beyond XenApp 6) but as far as XenApp' loosing it's crown as Citrix’s flagship product is concerned; I think the answer has to be that yes we are seeing a transition here. I’m not sure we will get to see the passing of the torch acted out on stage during next year’s Synergy keynote, but I can see a few moist eyes in the near future.

XenDesktop 4 will be available Nov. 16. Prices range from $75 for Standard Edition (XenDesktop only), through to $225 for Enterprise Edition (XenDesktop + XenApp) to $350 for Platinum Edition which includes full management and security capabilities. XenApp customers with Subscription Advantage will be able to trade up to XenDesktop 4 with discounts of up to 80 per cent. 

More later …

Is Citrix getting ready license HDX?

In order for virtual hosted desktops to succeed in business (note, that's business not IT) there are only three things that really matter, applications have to work, the user experience has to approximate that of a local desktop and it has to be cost justified.  We don't need to spend much time thinking about the first or last items, if it adds no business value and finishes up costing more than a managed local desktop we can't justify implementation, and we know that if an application works on a local desktop it will function just fine on a VHD desktop guest.  That only leaves user experience. While IT users understand and are willing accommodate an interface that is slow to respond when remoting into a server or desktop for  support or troubleshooting activities, we have to acknowledge that the experience can be frustrating. Offering such an experience to business users will rapidly lead to  unemployment. 

The yardstick that all remote application delivery solutions are measured against is clearly Citrix's HDX protocol. It's still not perfect, but we all know that it's pretty good. Regardless of the many marketing driven claims offered by its competitors Citrix has always maintained a comfortable lead in its ability to deliver the best user experience regardless of network bandwidth or latency. The problem with being the technology leader though, is that after a while it becomes increasing harder to innovate and maintain the size of your lead. Of late, alternatives to HDX have come to market that approach the user experience offered by HDX albeit with some compromises around platform (Teradici), low bandwidth/high latency performance, or ability to deliver multimedia content. So while Citrix still maintain some degree of advantage at this point the alternatives to ICA/HDX are becoming increasing attractive, and when Vmware's partnership with Teradici deliver's a software implementation of PCoIP  Citrix 'could' find themselves playing catch-up for the first time in their lives (VMware came on site a couple of weeks ago to demo the latest beta of soft PCoIP and while I'm not able to offer any comparison with HDX, it certainly appeared to be a viable technology). 

Faced with this pressure we have to expect some sort of response from Citrix and it's looking increasing likely that this response might be to make HDX available outside XenApp and XenDesktop. I looked at why Citrix might consider licensing HDX to other WAN Optimization Controller vendors in a previous post, this time let's look at why Citrix might consider other options for HDX as a stand alone display protocol.  

Citrix have an number of interesting options here, ranging from full open source release (not going to happen!) through to offering it for sale as a stand-alone product (possible, but not at the top of my list). Brian Madden made the point for Citrix releasing HDX as a no cost/low cost stand alone display protocol and while I agree that this is attractive I'm just not convinced that Citrix would be willing to take this large a step yet. What might be more likely is to consider licensing HDX to players in the emerging Cloud Computing DaaS market. DaaS service providers are acutely aware that there are few visible differentiators between offerings, with the inevitable latency present in Cloud Computing offering HDX could be a deciding factor in choosing between one DaaS provider over another. 

If Gartner are to be believed hosted VDI will account for $65 B by 2013. Citrix are smart enough to know that there is no way that XenDesktop will win all that business, but there's a real possibility that if they license HDX,  Citrix will tap into a much larger part of the market and create a  substantial new revenue stream for very little effort. I know that some people will say that Citrix would never license HDX, but if I remember correctly this wouldn't be the first time. I seem to recall that back in the early days of WinFrame Citrix licensed ICA for use in a version of PC Anywhere. So while I can't make any predictions about who Citrix might license HDX to, the prospect of a DaaS provider delivering DeskTone (or whatever) over HDX is, I think, within the bounds of possibility.  

Can we possibly expect Citrix to repeat their Project Independence desktop hypervisor announcement strategy and break this news next week to coincide with VMworld.