Microsoft: It's time to consummate the marriage with Citrix already

The relationship between Microsoft and Citrix is much like that of a man who has been serially dating the same woman for 15 years, but never seems to be able to consummate the deal. He's afraid of commitment, or perhaps like that old expression goes, why would he buy the cow when he can have the milk for free?

Click on the "Read the rest of this entry" link below for more.

For those of you not familiar with the long-lasting whirlwind romance, let us go back a bit, back when many of you readers were still bedwetting. In the early 1990's Citrix introduced its first product, MULTIUSER, which was a specially modified multi-user version of the OS/2 operating system, which for those of us old enough to remember was still a cooperative effort between IBM and Microsoft at the time. Its primary target market was for companies who wanted to provide remote access to applications from branch offices and feild personnel, which often only had 19.2K/28.8K/56K dial up modems or leased lines. This is the age when 3COM, Digi, and US Robotics were king, when BBSes networks such as FidoNET and Online Services such asCompuServe, BIX, and Prodigy were the predominant form of e-mail communication. AOL had only just been formed as a company, and  the Web itself was only in the early proposal stages. Broadband for the most part was science fiction for most companies let alone private citizens, and anyone that could afford full T-1 line(s) certainly was putting it to very good use, most likely splitting them up into dial up lines via ISDN multiplexing and voice carrier as well as for data. Citrix, with its very thin ICA remote display protocol, allowed companies to make the most of the bandwidth they had and still deploy rich client-server apps to remote offices and mobile workers.

When OS/2 no longer became a viable platform for Citrix, the company entered yet another technology licensing agreement with Microsoft, and created WinFrame, a multi-user version of Windows NT 3.51. Citrix continued to do a good business with WinFrame up until the mid 1990s, when they were forced to negotiate another license with Microsoft, who was then developing NT 4.0. As a result of these negotiations, Citrix agreed to build an add-on product to NT 4.0, MetaFrame, which would provide the enhanced multi-user and load balancing/clustering functionality with their ICA remote display protocol and forgo development of its own modified OS. In return, Microsoft would get much of Citrix's technology built-into NT and Windows 200x , as Windows Terminal Services.  To differentiate from MetaFrame, Terminal Server would utilize the less-optimized RDP protocol rather than Citrix's snappier and more adaptive ICA, and would have no load balancing or advanced application publishing features, such as for Web-based Windows application portals.

This relationship going forward between Microsoft and Citrix was strenuous at best and became a huge hassle for the corporate IT decision maker considering a MetaFrame purchase. You see, in order to properly license the use of Citrix in a large corporation, you had to have a Windows Terminal Server license or equivalent Windows Desktop license for each concurrent remote desktop session, and then on top of that, you needed to buy Citrix connection licenses for each client actually connecting using the ICA protocol -- a practice that I still like to call the "Citrix double whammy". For many organizations, this was and continues to be a deal breaker. Many companies chose to use Windows Terminal Server instead, because they already had volume licensing agreements with Microsoft, and unless they really needed the load balancing functionality and were seriously bandwidth constrained across their WANs, there was no need for MetaFrame at all. Many organizations adapted a hybrid strategy, where LAN-based and less bandwidth constrained users would use Terminal Services and a smaller, global, remote user base would use MetaFrame, in order to reduce their Citrix licensing overhead.  Still, many of us wondered why at this point in Citrix's history when MetaFrame was created why Microsoft just didn't go and buy the company in the first place -- it would have made it so much easier for their customers. MetaFrame eventually became Citrix Presentation Server, and eventually Citrix XenApp. Still, the "double whammy" persisted, no matter what the product end up being called.

Set the dial on the Flux Capacitor to October of 2007. Citrix completes its purchase of XenSource, the company behind the development of the open source Xen hypervisor. With XenSource, not only is Citrix now a major player in Thin Client application deployment, but also in virtualization. To make matters more complicated, XenSource is also responsible for the technical architecture behind Microsoft's upcoming bare-metal "Viridian" hypervisor, which later became known as Hyper-V and is included in Windows Server 2008. It also designed and created the Hypercall Adapter which allows Linux operating systems to run fully accelerated within Hyper-V. In September of 2007, Citrix and Microsoft commit to standardizing on the VHD file format for virtual machines, so that Citrix XenServer and Hyper-V will easily interoperate with each other.

One would think that with such key technology of Microsoft's being based upon the technical efforts of XenSource, that the Redmond giant would have finally done the deal and consummated the marriage with Citrix. But no. Citrix remained a spurned woman, left to compete with VMWare, with its 80 percent market share. This didn't stop Citrix from slogging it out on its own and developing some interesting products though, such as XenDesktop, which merges the thin client capabilities of MetaFrame's ICA protocol with the XenServer virtualization platform. While it is technically at parity or even superior to VMWare's VDI for desktop virtualization in some respects, it hasn't gotten a lot of takers so far.

Flash forward one year later to September, 2008. Red Hat shocks many in the Open Source and virtualization communities with its $107 million dollar purchase of Qumranet, an Israeli start-up formed by one of XenSource's founders. Qumranet brings to the table the open source KVM hypervisor, as well as a high-performance thin-client protocol that gives Citrix's ICA a run for its money and then some, as well as a remote desktop provisioning platform that rivals both XenDesktop and VMWare VDI. Suddenly, Red Hat is poised to challenge both Citrix and VMWare in desktop virtualization. Now, guess who recently released a new high--performance bare metal hypervisor and has no desktop virtualization solution at all? Do you all see where this is leading?

To battle VMWare and Red Hat, Microsoft and Citrix need to finally consummate the marriage. One, so that the Hyper-V and Xen technology can be more quickly integrated, and so that Microsoft has a desktop virtualization platform for Hyper-V. Second, as Red Hat has disclosed to me that they intend to Open Source the SPICE protocol as well as large portions of the SolidICE product, the "double whammy" is going to become a big issue for Citrix and Microsoft going forward -- if XenApp became part of Windows Server and XenDesktop was integrated into System Center Virtual Machine Manager, and if the "double whammy" could be reduced simply to a "single whammy", Citrix and XenDesktop adoption would increase significantly and have a fighting chance against Red Hat's virtualization technology.

But if Microsoft bought Citrix, what of XenServer and the Open Source Xen Project? Well, there are two possible outcomes for this. One, would be for Microsoft to continue to maintain and sell the XenServer product and fully integrate it into its System Center product line, and to maintain Xen as its first major Open Source project -- a practice it has little experience with but could give them the desired Open Source street creds that it currently lacks. But there is another answer.

Sell XenServer and the Xen Project to Novell for a nominal fee, with rights to use any XenServer code or previously licensed intellectual property that is used in Hyper-V until a contractually agreed upon date, or manage the Xen project with Novell as an extension to their technology sharing and interoperability alliance. This way, Novell finally gets a hypervisor to call its own, and will allow it to complete their own enterprise virtualization platform stack with their Platespin acquisition from March of this year, and Microsoft will get to reap the benefits of having a partner that can manage a major Open Source project that will ensure compatability with their own solutions going forward.

Is it time for Microsoft to do the honorable thing and buy Citrix? Talk Back and let me know.