The day after a computer glitch resulted in a seventy minute period when no one seemed to know where the Dow Jones Industrial Average was headed, only to see it drop nearly 200 points in a matter of seconds once reporting was resumed,I was having a much better day than the IT folks at Dow Jones and the New York Stock Exchange.
While those poor IT souls were doubtlessly trying to figure out just what went wrong, I was having lunch with Bill Coleman, the co-founder of BEA Systems and current founder of Cassatt Corp. While most of Silicon Valley seems to be enamored with the idea of creating the next big social networking scheme, Coleman's company is geared to helping companies manage their IT systems. Far less sexy than the next Facebook, but far more important to someone who would actually like to get a return on investment for all those IT dollars.
I asked Bill what he thought about the Dow glitch. I suppose glitch is too light a word for a mistake that may have erased a couple of billion in stock value as trading indexes tied to the Dow suddenly had no Dow to guide them. He said there were two avenues a CIO can take when their system investments go weak at the knees.
One, they can go back to the big hardware vendors with their tale of woe. The big system vendors will be glad to sell them an overprovisioned system that they will claim will carry them through any emergency. Should that emergency take place and the systems fail, then the response will be to sell the customer even more systems. "But once the system reaches full capacity, that's it," Coleman told me over lunch in Boston's financial district.
In that approach, rather than looking for the weakest link in a chain, you sell the customer an entire new chain. And should that fail, you sell them yet another heavier gauge chain.
The other approach, and yes the one Cassatt is selling, would be to create a system that automatically looks for the weakest link and directs computing resources towards bolstering that link. This is not predictive computing, but applying computing resources in real time to real problems. This approach is a fundamentally different approach to computing and requires a system that is always aware of its resources and can direct resources far faster than any human operator could ever respond.
I am sure there will be lots of finger pointing over the next couple months as both Dow Jones and the NYSE try to figure out what went wrong. I'm not sure what systems Dow runs on. The NYSE is, I'm fairly sure and I'm looking for clarification, an IBM system called TradeWorks which was promoted under the claim of "Extreme Availability."
The interconnection of the Dow Systems and the NYSE and, moreover, the interconnection of all worldwide electronic trading systems doubtlessly offers new avenues for faster trading and for fortunes to made within moments. However, those fortunes can also be lost within moments. Earlier this week, traders went back to paper and posting trades on whiteboards in an attempt to keep the system moving. Before you decide to buy a new heavier chain to keep your IT system anchored, you might want to think about an approach that constantly inspects the strength of your current chain and in real time fixes the links that are faltering.