John Katsaros

The Fourth Cloud

Mark Templeton's keynote at Synergy, the annual Citrix users' conference is usually pretty interesting and this year was no exception. Templeton introduced the "Fourth Cloud" — the Personal Cloud — to what has largely been a three cloud world (Public, Private and Hybrid). The Fourth Cloud is an important concept because it defines the place where corporate IT meets the workforce. If IT gets this right, they'll be able to minimize the footprint of what IT has to own and what applications IT has to manage. Templeton's point is that it is in IT's best interests to make this happen since they'll ultimately benefit. He pointed out three essential ingredients for the personal cloud — 1) Context Switching — any to any connectivity, secure where necessary; 2) Self Service (including an apps store and of course things like Google Apps); and 3) Identity and preferences — managing the convergence of personal and work use. Citrix's view of the personal cloud is that it has: A front door that orchestrates delivery, Follow Me apps and data, Authentication /Security (Single Sign On, password management...), Cloud Connectors (for both business cloud and social cloud interactions) and Management (activity, Access Control, SLAs...). The Personal Cloud is a great idea — rentable applications, disposable devices, automatic synchronization, self service, and bridging into consumer apps — offloading from IT functionality and cost that is better acquired as consumer products. At previous Synergy keynotes, Templeton shined a light on the concept of BYOPC (Bring Your Own PC) which eventually has become a familiar phrase. The idea of a Personal Cloud (and IT's relationship to it) fits in nicely and takes BYOPC to the next level.

Zynga Scale

Also at the Citrix Synergy conference Simon Crosby introduced a mystery speaker who turned out to be Allan Leinwand, Zynga's Chief Technology Officer — Infrastructure who told a fascinating story of scale. Zynga's games include Farmville and Words With Friends. The demand on Zynga's infrastructure is amazing — 250 million monthly active users, five new players join Zynga every second, over 10% of the world's Internet population plays Zynga games every month. To make things really difficult, it's hard to predict how fast a new game might grow. Let's say they launch a new game on a Monday. They have no idea what the demand will be on Friday. If the game is a flop, they won't need much infrastructure. If it's a success and goes viral, the demand on the infrastructure could be huge — Words With Friends processes 1,000 moves a second. Early on the company couldn't grow its infrastructure (space, cooling, servers, networking equipment and transit) fast enough. The scaling is amazing - for every one server they had in January 09 they had 75 just two years later. The straw that broke the camel's back was Farmville which in 2009 grew to 25 million daily active users in just five months — a growth rate that was impossible for them to grow out their physical infrastructure that quickly — so they made a quick move to start using Amazon's AWS. Realizing they were trading between operational expense and capital expense they decided to build their own private cloud which they call zcloud — fast scaling, open source, Xen, X86 architecture — which uses RightScale to provision services on both their private cloud and Amazon. Zynga's growth is an interesting example of how quickly things have changed in the infrastructure world on both the demand size (supporting users) and the supply side (providing server cycles). This would have been a completely different story five years ago with the outcome costing hundreds of millions of dollars to provision infrastructure ahead of demand. Who knows, maybe their next game is going to be "infrastructure scaling with friends" for all of us infrastructure geeks to play around with application demand and server availability (Humm, that's not such a bad idea...)

If you have to train someone to use it, don't build it

We probably need to create a category of "Templeton-isums" to keep a list of some of Citrix's Mark Templeton's sayings. What Yogi Berra's Yogi-isums ("Nobody goes there anymore, it's too crowded") were to baseball Templeton-isums are to IT. The latest that came from his Synergy 11 keynote were "don't fight it, feature it" referring to IT's reluctance to unleash iPhones and iPads onto the corporate network. And "Never do an IT project that requires training" — great advice. Over the last few years while you think we've made progress the truth is that the Internet has turned us into telephone operators, travel agents, bank tellers etc. There's no way that you can train all the users to know how to use these apps — they have to be well designed. It's the simplicity of these "isums" makes them memorable.

They Eat Their Young

Microsoft has a reputation for being a developer's best friend as the Microsoft platforms are successful in larger part due to the support of their eco-system. There are many examples - SharePoint with several hundred partners and Microsoft's relationship with Citrix. VMware on the other hand doesn't have the same magic touch developing its ecosystem, in large part due to its product line expansion strategy. Probably the best way to think about being a VMware partner is that VMware will develop 80% of the functionality itself and take down 80% of the market opportunities unfortunately leaving partners with a much smaller and technically challenging market slice. At VMWorld there were new examples of this as VMware introduced vShield and in doing so changed the bar for companies trying to add security functionality to the VMware platform - setting the stage for VMware to roll up the bulk of opportunities in this space. And two announced acquisitions, single sign on vendor TriCipher and its $100M acquisition of Integrien for visibility must have taken the wind out of the sails of a dozen companies providing the similar capabilities. VMware is on top of its game so if its ecosystem rules are 80/20 then that's the way the games going to be played this time around.

What a Difference a Year Makes

At the end of August Citrix announced its plans to acquire VMLogix, a company focused on virtual datacenter service such as self-service. In July, Quest announced its plans to acquire Surgient which was largely in the same space. And of course way back in 2006 VMware acquired Akimbi which some people say formed the roots of project Redwood which launched at VMWorld as vCloud Director. While a year ago these products were focused on providing services to the test/dev world, today the focus has changed and they're viewed as the initial service platforms for service providers bringing virtual datacenter functionality on line.  Interesting how this category changed so quickly.

Citrix Synergy 2010

The Citrix customer conference was local (for us) this year. The meeting has bounced back nicely from the receded form in the recession and had sold out. Citrix meetings are always fun to attend: the user group has an enthusiasm and loyalty I've only seen before at things like the Apple developer conference (where the Cool Aid flows like wine). And we know the management team reasonably well and are always impressed by their collective cheerful hard-driving spirit. 2 1/2 years ago Citrix, when it acquired Xen, made a huge ($500M) bet on virtualization of various forms as a way of breaking out of (and defending) their core application delivery business. They've had plans to more thoroughly integrate all the pieces and now this all seems to be coming to pass. In the past, Mark Templeton's keynotes were best understood by the Presentation Server faithful, but that wasn't the case this year at all. Mark's opening keynote (replete with demos) paints Citrix' vision very clearly — how we get to our stuff from anywhere and any device, in an ever more wireless and mobile world. It takes a lot of technology and careful engineering (and re-engineering) to make it all happen but in the end the vision is simple and quite compelling. Watching Mark's keynote on Web replay is highly recommended.

Dell's Turn at the Podium

Michael Dell was the keynote speaker on the second day of the Citrix Synergy conference. You might wonder why he even chose this venue to talk considering that much of the work that Citrix is doing in desktop virtualization eventually ends up reducing PC differentiation — for instance, with a "bare metal" hypervisor it's hard to tell the difference between a Dell laptop or a Lenovo since it's all virtualized. Although he wasn't asked that question directly, he had a pretty good response to the subject along the lines of "When something comes along that customers want you stand in the way of it at your own peril." He made the observation that eventually the complete product offerings for the coming generation of virtual endpoints will have an all encompassing solutions including servers, storage and services. Having bought two out of three of those pieces recently — storage focused EqualLogic in Nov, 2007 and services focused Perot Systems in Sept, 2009 — he clearly understands the importance of building an alternative strategy as virtualization reduces his core differentiation. It will be interesting to see how all of this plays out.

Fanboy

We're not trying to be Simon Crosby fanboys but it is sure fun to watch Simon (Citrix's CTO of the Data Center and Cloud Division) in a panel discussion because he clearly explains complex concepts and put misconceptions to rest. This week during an Interop keynote panel discussion Simon took on the notion that enterprise private clouds will only be housed on enterprise locations by pointing to the IRS (which already has much of its data housed in third party data centers and to service providers like Terremark which, by specializing in managed services, houses its customers' data in locations that are much more secure than their customers could ever match and protect these locations with armed guards). Simple, easy to understand and convincing. The Interop Panel Keynote can be found here.

Getting Our Heads Out of the Clouds

A lot of early discussion on Cloud computing didn't make much sense in the cruel light of day: Cloud computing is unlikely to be cheaper than doing utility computing in your own data center; Cloud Bursting has to take into consideration the laws of physics and reality of moving large volumes of data. On the other hand, the underlying forces couldn't be more important: the enablement of utility (on demand) computing models; the ability to evolve service-oriented IT; the enablement of software/platform and infrastructure as a service.  In our after-dinner discussion we'll attack these interesting issues and try and illuminate the key points (or at least have entertaining fun in the process!). Peter Christy will lead the discussion as always. He'll be joined by Citrix product executive Sunil Potti who is right in the middle of their Cloud efforts, and Scott Johnson, Venture Partner at Alloy Ventures (our generous sponsor for the dinner).

Citrix Ups their Game

Citrix went all out at their Synergy user conference last month in Las Vegas with a fistful of announcements centered around virtualization and cloud computing - 1) Enhancing its C3 Cloud Computing Platform for apps and desktops to be delivered as a cloud service, 2) The C3 Lab built on Amazon Web Services, 3) Citrix Essentials (virtualization management) upgrades to its XenServer and Hyper-V product lines and 4) Changes to streamline licensing to make it easier for the Citrix installed base to migrate to the new platform.  These are big steps in its campaign to unseat VMware as the virtualization incumbent.  Either Citrix has been having difficulties gaining ground and they've had to up their game or Citrix sees good growth opportunities with its Cloud strategy and focus on virtualization management — or perhaps both?