IT Opinion Piece:
A lot of us have been viewing the Google deployment of cloud services in Los Angeles as a repeat of what killed Netscape: Over promising a technology that couldn’t deliver. In my view, it hasn’t been going well. So poorly, in fact, that Google last week implied that the LA effort was impossible. I spoke with Richard Plane from Harris last Friday to understand why they are able to do what Google can’t. You see, Harris provides a cloud solution that is fully compliant with financial, healthcare, and government standards and they got there by designing compliance in from the beginning.
Google vs. Netscape vs. Microsoft
All three companies initially approached the enterprise in the same fashion, with products that were cobbled together from acquisitions of technologies that, individually, weren’t compliant. Netscape was almost painful to watch as my team had recommended against it in one of our largest clients and they had decided to go with it anyway. It never worked properly and Netscape eventually failed as a company and virtually all of the folks we had been working with at our client company got laid off or were managed out of their firm as a result. This showcased the very real risk of working with a company that fundamentally couldn’t make the solution work.
Now Microsoft, in the late 90s, was in the same position and they had a huge failed event called “Scalability Day” where they attempted to prove they were enterprise ready. However, even though they weren’t ready there was a key difference. One of their key accounts had replaced a mainframe with Windows NT servers successfully, however when called he said something to the effect that “yes it worked but had we had to pay for this solution we could have bought 3 mainframes.” In short even though it didn’t work, Microsoft had funded the effort to success even though it cost them an estimated 4x what they charged for it. It was a massive loss but the client was kept whole and it is through decisions like this that they became an enterprise class vendor. I should point out they never admitted that what they were attempting was impossible and eventually it wasn’t, though it took nearly another five years to get there.
In my opinion, Google’s seeming admission that what they are attempting is impossible basically invalidates them as a vendor in this class and that could have been avoided by following Microsoft’s lead, or by actually having an enterprise solution. They could have bought one and I’m kind of surprised LA didn’t make them.
Unlike Google, Harris built their enterprise solution from components that were enterprise ready. This first step was likely the most critical.
BMC Cloud Planning and Design: Any enterprise project should have at the front end a solid plan; BMC has developed what may be the most advanced and, more importantly, hardware-independent planning tool. This tool, at the core of which is a training workshop, also contains their Atrium Discovery and Dependency Mapping, and Capacity tools in a bundle. This allowed Harris to define a set of requirements that would create a system that could handle 200,000+ secure virtual machines all compliant with the various enterprise, finance, and government security requirements. It defined, at the front end, without any predisposition to any one hardware vendor, what would be required to make the solution work. This is likely where Google failed, they didn’t adequately define the problem /solution so what they created didn’t work.
VBLOCKS: A VBLOCK is a unique product created through a partnership between Cisco, VMware, and EMC that was designed from the ground up using the enterprise technologies of all three companies to create a cloud platform that would both massively scale and be fully security compliant. It really is the only effort of its type at this scale that crosses as deeply as three enterprise vendors. As it turned out, the Cisco component was one of the most critical and Cisco was engaged to create a network barrier between the hosted VMs so that security compliance could be assured.
A few decades ago there, at a political event, I watched a politician argue that for the cost of one battle tank you could equip hundreds of Chrysler Cordobas with big guns (that is scary is it appears someone may actually make a conversion kit) and overwhelm an enemy much more cheaply. He was half joking because the end result would have likely been something like the Porsche Tank in Lenard Part 6 (the worst movie that Bill Cosby ever made). The Porsche Tank, and maybe the entire movie, is kind of a visible representation of why Google and Netscape failed (though it is creating some interesting mental images of what Microsoft would have had to create to build a successful one). In the end, though, it also points to the core tenant, if you build a solution from compliant parts the solution will probably be compliant, if you don’t it won’t and even if the firm is willing to pay to make it work you’d be better off on another path.