You picked Azure as number one, AWS as number two, and then eventually GCP as number three. If you had to guess what the result would have been, how did the actuality match up to it? What was the rationale for those choices?

Everybody was used to the procurement process; I learned a few lessons that day. I had always worked with customers, as a vendor but when you're a customer – depending on what institution you're working with, and what kind of company you're working with – you can't win all battles. You're going to pick the fights you want to win. This one was not worth picking a fight. All else held equal there, more or less at the end of the day; more or less the same few if you work around some of the issues. For me, I would have picked Azure if it was all my perception, not an all-Microsoft show. My guesswork is Microsoft knows how to work with procurement processes better than anybody else. To me, that is legal documents, all of the T&Cs, all the paraphernalia that everybody else takes for granted. They shine there. Plus, it does help that Satya shows up in the deal, personally, because the number is big enough. This is classic; I've seen this all the way from Tibco onwards; sales guys usually don't want the CEO to show up in the deal because a lot gets given away.

What does in-memory mean?

Usually what happens is, databases store their records on disk, and just fetch the basic core amount or handful of you to operate on and then they keep cycling through that and keep fetching. With in-memory, let's maybe simply say that handful has become almost like an infinite field, in some ways. As far as the consumer or the user of the technology is concerned, performance changes drastically. You don't have the issue of fetching from disk and getting the reads and the fetches don't take that long. Even a nanosecond is too slow. It's like saying you have the letters laying right in front of you, versus you have to go to the post box to get it versus post office to get it, different levels.