Sunday, August 30, 2009

Citrix Set to Announce an Open-Source vCloud on VMware's Turf

Word is trickling out that Citrix plans to announce on Monday, at the kickoff of the VMworld conference, an open-source version of VMware's vCloud called the Xen Cloud Platform (XCP) that's virtual machine agnostic, non-proprietary, and significantly less expensive than vSphere 4 -- all with the promise of no vendor lock-in.

The Xen Cloud Platform will supposedly surround the Xen hypervisor with a complete runtime virtual infrastructure platform that virtualizes all server, storage, and network resources. In addition, it will be completely
virtual machine agnostic, in contrast with VMware's vSphere platform which currently runs exclusively on its own infrastructure.

In an interview on Friday, Simon Crosby, CTO of Citrix's Virtualization and Management Division, hinted at "substantial contributions" to come from Oracle, HP, and Novell.

I can hardly wait until Monday. It's sure to be an interesting week in San Francisco, particularly if you're at the VMworld conference.

Tuesday, August 25, 2009

The Cloud: The Sky is Not the Limit, The Network Is

The inherent mobility benefit you get from a virtualized infrastructure is driving IT shops, large and small, to take virtualization to the next level to begin to deliver dynamic IT services to the business. At the server and storage layers, we've seen a lot of innovation and tight integration between the various virtualization software and storage vendors. However, at the network layer, the integration gets much trickier to truly deliver on the promise of virtual machine mobility. Today, key network settings like QoS, ACLs and VLANs only allow policies to be applied at the physical port level. Whereas in a virtual environment, policies need to be applied to workloads at the virtual machine layer to enable networks to extend out to the VM boundary.

No doubt, the next big innovation coming is in the network layer in order to gain visibility at the upper layers of the software stack, and automate core network services that are largely manual, and labor-intensive today.

Cisco is ahead of the curve in this regard. They have worked closely with VMware on VN-Link to begin to deliver on the promise of VM-aware networking. I suspect we'll soon be hearing more from other established players like Brocade, Juniper, F5, and networking start-ups such as Arista Networks.

Thursday, August 20, 2009

Don't Get Sucked into The Cloud Hype Vortex

I've been in the industry long enough to know better than to get sucked into the hype vortex of vendor marketing run-a-muck, complete with declarations of major paradigm shifts, self-healing architectures, and newfangled delivery and deployment models of computing. After all, I've been on the front lines selling client-network computing, client/server computing, Internet computing, grid computing, etc. over the years. And I'm painfully aware of the wide gap between when customers write checks to actually adopt new technology, and when vendors promoting the technology predict they'll to be adopted.

Yesterday, a colleague and I went to a full-day VMware vSphere 4 SolutionTrack technical sales training. Apart from VMware Marketing having a bad habit of changing the names of its products every few months and confusing partners and customers alike, the hype the company is spewing about the Cloud, and its vSphere 4 platform being the first Cloud OS, has become deafening. Frankly, I believe it will ultimately do more harm than good in enhancing the company's credibility in its future combat with Microsoft, Oracle and IBM.

I do admire VMware in many ways. They employ a lot of scary smart people, and have a bold vision around "IT as a service" that I find compelling. And their vision gets far more coverage than that from Microsoft, Oracle, and IBM... at least up to now. But, I keep lamenting why VMware doesn't refrain from so much hype, and simply present its vision in the context of the next generation data center that builds on numerous computing models that came before it, providing transformational benefits to IT over the long term.

At the end of yesterday's session, I gave the trainer my suggestion that he and his counterparts would be much better served by articulating VMware's cloud vision around
words and key positioning statements like:

* Evolutionary
* Evolving
* Comprising m
any phases that will take years to fully realize all the benefits
* V
irtualization is a key underpinning for building cloud infrastructures
* It's relevant to you "Mr. Customer" whether your long term objective is to build a private cloud, public cloud, or a hybrid cloud to best align with your changing business needs

I seriously doubt the VMware trainer gave
any credence to my suggestions. I think he's been sucked into the hype vortex. But based on what I learned yesterday, I'm more convinced than ever that we're a long way from seeing any vendor take its vision and be able to put all the essential components on a cloud. Until an industry ecosystem achieves openness, interoperability, self-provisioning, elasticity, programmability, trust, control, etc., I won't be contributing to the hype around cloud computing.

Sunday, August 16, 2009

Can VMware or Oracle Do for Java What Sun was Unable to Do?

Back in August 1999, Sun Microsystems President Ed Zander spoke at a New York press conference where Sun unveiled its strategy for using StarOffice as the centerpiece of building what Zander referred to as a "service-driven network". He declared, "We're seeing a second generation on the Internet... In 1995, it was about the network, in 1999, it's about services." At the time, Sun hoped to encourage a new generation of developers to create web-based applications using StarOffice open source code. Today the battle is in the cloud where Microsoft is countering the Google Apps lineup.

In 2001, Sun opened a new chapter in Java history to more effectively compete with
Microsoft's .Net/SOAP, declaring the Java framework one of the key underpinnings of a service-driven network.

As was often the case throughout Sun's history, it was prescient in its vision, but allowed its competitors to capitalize on its vision with a keener sense of timing, superior marketing, and business execution.

VMware/SpringSource is opening yet another chapter in Java history, envisioning an environment in which developers can define not only how objects connect with one another, but
how they should be packaged into virtual machines and deployed into a virtualized infrastructure.

.Net and the Microsoft tool set are already quite capable of delivering on this similar vision. And to keep the coming batttles even more interesting, Oracle will soon inherit the stewardship of the Java platform, and it's a good bet it will do a better job than Sun.

Time will tell who can best capitalize on its vision with developer mindshare, marketing, and execution.

Tuesday, August 11, 2009

Will VMware/SpringSource be the Future Foundation for App Development?

We've heard many companies over the last decade hype their vision of a post-OS future whereby the operating system is rendered irrelevant. Back in the mid-late 1990's, Sun Microsystems, Netscape, Novell, and others declared Java the game-changing technology that would be the beginning of the end for Microsoft. We saw how that strategy turned out for Netscape and Novell (and Sun for different reasons), as Microsoft eventually cut off their air supply by bundling its "good enough" browser, directory server, and other widgets into the Microsoft stack for free.

Today, the "new" Netscape and Novell is VMware. Clearly, VMware CEO, Paul Maritz, a former Microsoft executive himself, understands the stakes and the formidable competitor he's up against for dominance in the data center. That's why he and other smart minds at VMware figure it's well worth paying $420 million to purchase SpringSource who had roughly $20 million in annual revenue. It's an expensive bet, but VMware recognizes its strategic vulnerabilities. And it's keenly aware that if it simply continues to innovate at the hypervisor layer, it's only a matter of time before it's driven into irrelevancy by the folks in Redmond -- Similar to what Microsoft did to Netscape and Novell nearly a decade ago.

With SpringSource, VMware gains upper layers of the software stack and the ability to decouple the application from the operating system in much the same way it decouples the operating system from the physical hardware infrastructure today. If Maritz has his way, enterprise developers won't write their applications to an OS anymore. Instead, they will architect applications using the SpringSource open source framework and Java Apache Tomcat application server.

It's a compelling vision, but one fraught with substantial risks. On one side, VMware has Microsoft with Windows Server 2008 R2 and Hyper-V releasing this Fall at 1/6 the cost of VMware vSphere 4. On the other, VMware faces new rivals in 800 pound gorillas, IBM and Oracle, who both happen to own well established Java development platforms in addition to owning the back-end databases and in the case of Oracle, the enterprise applications that run on them. And most of the mission critical middleware and data-centric applications have yet to be virtualized in production by the vast majority of enterprise customers.

Probably the biggest immediate loser in VMware's SpringSource acquisition is Red Hat. It's revenue streams remain largely tied to Linux OS subscriptions and support and they haven't seemed to be able to turn the JBoss acquisition into a sustainable advantage.

If history is any indicator, VMware's odds are not very good against the likes of Microsoft, IBM, and Oracle. Maritz will need more than the firepower of SpringSource to make VMware the future foundation on which enterprise applications are deployed.

Saturday, August 8, 2009

The Cloud: Next Generation "Just-In-Time" Computing

When first developed by the Japanese in the 1970's, the idea of just-in-time (JIT) marked a radical new approach to manufacturing. It cut waste by supplying parts only as required. The old approach became known as just-in-case where an inventory of parts was held for every possible eventuality...just in case. Throughout the 1980's and early 1990's, JIT became a business philosophy and a system for production management; However, there was no real master plan or blueprint for JIT.

After attending EMC Forum in Long Beach, CA last Thursday, where VMware, Cisco, and EMC (referred by EMC as the 'VCE Alliance') got on stage to talk about a shared vision around cloud computing, it occurred to me this next generation computing model correlates with JIT. After all, the goals of cloud computing are pretty similar. Cloud computing is all about cutting waste to drive efficiency so IT doesn't continue in its old ways of over-provisioning resources, just in case there's an unforeseen spike in demand. And just like JIT had no masterplan or "one size fits all" approach to follow, there's no one cloud computing model that's the de facto model to embrace over another -- at least not now or anytime soon.

As the saying goes,"Whatever is old is new again". Cloud computing, although transformational in many ways, builds on established trends that came before it -- network computing, pay-per-use, software-as-a-service (SaaS), and just-in-time manufacturing.