Sitting in my cubicle today I heard this over the divider, "What it is, is what it is." Meaning the data warehouse data load job was running longer than expected and no one could explain why. But that was okay because in the computer business we have to expect this sort of thing. Delays are delays.
It reminded me of a sign I saw in a Canadian Pacific Airlines ticket counter one day, back before that company went bankrupt. The sign said,
"In order to serve you better, we are improving our computer environment.
Thank you for your patience!"
At first it struck me funny that I was to expect to receive lousy service while computers were being upgraded. Then I wondered if service would actually get better after the upgrade. Well, given that the company went out of business soon after that, I guess service got worse.
What it is, is what it is. Live with it. Or not…
I think my systems provider should deliver a system where "What it is, is what is expected." Of course the only way to know what to expect is to test the product prior to giving it to the customer. And by test I mean full functional tests, user acceptance tests and performance and volume test.
The challenge in the Big Data world of Data Warehousing is people forget that it takes time to process a billion rows of data. And every effort to build a system that can handle the large volumes needs to be made up front in the design and development stages, prior to bulk data loading and even prior to the performance & volume tests.
There is one school of thought that says, first make it work, then tune for performance. The challenge with this is, what happens if you tune all the performance you can out of your current design and the thing still runs like a tortoise. Then it’s back to the drawing board.
Somewhere else on this blog site there’s an entry call "Performance is Always an Issue". It’s my mantra.
What it is, could be better.