A couple years ago I was ready to quit private contracting and subcontracting. I wanted to join a big deal company that really understood financial data warehousing. I had two choices, and ironically they were looking at me too. They were Beacon Analytics / Answerthink and Knightsbridge. I had worked with Knightsbridge in a sales capacity and I found them to be a very tight organization. They had people who worked with Embarcadero Technologies and they were seriously well managed - enough to walk away from crazy hungry claims of the sort that sink many services companies. I decided on Answerthink and soon after I recall that Knightsbridge was sold to HP.
The news that Knightsbridge went to HP was somewhat strange. What HP did have was a very aggressive attitude towards hosting all the hot enterprise technologies, including and especially the Hyperion suite. What they didn't have was a track record as being a company that could be counted on to deliver all that. Perhaps their purchase indicated that they were ready to change all that.
It had been my experience that most customers who had committed to HP hardware, especially those on HPUX were woefully behind in OLAP and price-performance. Expertise with Essbase and HPUX was rare indeed. After a time, IBM was the leader in Unix hosting of Hyperion apps and Essbase, whereas once upon a time, Essbase on Sun was the killer app. HP was about tied with Redhat from what I could see in the field, which meant essentially that it was an academic release.
Part of the reason was that Hyperion had pursued, for a short time, a relationship with the engineering support organization of Sun. They must have done something similar when it came to IBM after I departed. What that comprised was a deal whereby Hyperion developers could get up to date hardware, compilers and advice on how to develop for the OS kernels, memory models and various subsystems particular to the OS/hardware combinations. Who knows if such things go on today.
By the time I had been at Answerthink for an appreciable period, I came across HP again. This time going great guns about 64 bit computing. It was something and still continues to be something to get very excited about. What I've personally been able to accomplish on the HP Superdome practically defies description in the ordinary world of Hyperion apps. For a short time, we had some very interesting momentum. Such momentum is very difficult to sustain without a real paying customer, but a company like HP can afford to keep BizDev folks buzzing for a good long time, as well as keep configuration engineers setting up performance laboratories that make the case for blistering performance. The trick is maintaining the mix with a consulting organization who has a big customer attention, meaning bucks.
This is clearly what EDS provides. And I rather like the idea that HP is doing this the technical way rather than IBM's way of the brownbagging of Global Services plus the high concept dinner of PWC. The only equivalent thing left would be a Sun Micro acquisition of CSC. Then what's left? Perhaps Google acquires Tata? SAP acquires Infosys? That would be a shocker. But it gives me a clue about what could also be going on here. Like it or not corporate IT has failed. What have they failed to do? They have failed to keep Salesforce and Tata out of their business. They have failed to deal with grid computing, and they have failed to get their datacenters and moreover their global applications functioning as well as the web providers of the world have. There is a huge gap between what the largest companies like Boeing can muster in their IT and what modestly successful websites like The Washington Post online can do. The very idea of serving up enterprise data to more than 10,000 simultaneous users gives most IT organizations a severe case of cramps. If you can't do that in the online world, you're lower than toast.
So how do you ramp up internal corporate IT and enterprise data to the sub-second million-user scale of external websites? It may be apples and oranges unless and until you consider what I've seen in HP's labs, which is several hundered Essbase cubes running simultaneously. What kind of collaboration is necessary to get the corporate accounting data running in a 64bit cloud? That's the kind of thing we should begin to see if and when HP consumes EDS (or Google + Salesforce + Infosys happens). See?
It's time to take the lessons of Web 2.0 and even 1.5 and apply them to corporate IT 1.0. It's time for internal enterprise applications to run faster and be more integrated using some of the same technology that makes Google a faster way to get your company's 10Q data than searching through shared drives in your Microsoft Domain. Sound like SAP? Sound like Oracle? Yep. Only large cross functional services organizations teamed up with serious compute clouds and grids can do that. If that's what HP has in mind with EDS, it's going to be a brave new world.