May 16 2011
e = mc(imc)2
The opening portion of the keynote was a dramatic (perhaps overly dramatic?) introduction by Gabriel Byrne of Usual Suspects fame (among others). His talk was interesting but what I liked the best was his update on Einstein’s famous formula. It looked like this:
e = mc(imc)2
In this formula, “m” stands for mobile, “c” stands for cloud, and “imc” is short for in memory computing. The concept is simple. By leveraging the power of in memory analytics, we can set up information resources in the cloud that are consumed by mobile devices. The three technologies individually are strong, but together they become greater than the sum of the parts. One of the primary restrictions about mobile devices is their relative lack of computing power. (Later in the keynote one panel member observed that today’s smartphones do, however, have more computing power than the systems NASA used in the 60’s to land man on the moon. Can you imagine piloting your lunar lander with your iPhone? Wonder if AT&T has cell coverage on the moon…) Because of the lack of computing power, mobile devices are mostly consumption portals rather than calculation engines. That’s where in memory comes into play. By pushing the analytics back onto the server and hosting them in memory, a mobile device doesn’t have to be brawny, it just needs enough brains to connect and render.
Putting the systems into the cloud just ads ubiquity to the system. It’s no longer behind a corporate firewall, it’s available to anyone in your enterprise, no matter where they are. It’s a compelling vision.
The keynote was followed by a discussion panel which was both entertaining and thought-provoking at the same time. It was more about where we will all be in 30 years than the more immediate concerns of in memory computing. 🙂 I will post more on the panel discussion later today. If you can, try seeing if the session was recorded and can be played back online. It’s worth watching.
Dave,
I am not convinced that in-memory is any different from a RDBMS/disk array architecture in terms of what computing power the client requires? The client just waits for the answer either way surely?
Sure, in-memory may give a faster answer, but, does it effect the client computing requirements? I can’t see that.
regards
Simon
The latest iPad (and presumably the next iPhone) has more computing horsepower than a desktop PC had in the late ’90s. According to what I could find online, Apple’s A5 processor pulls 168.90 Mflops / s. These devices also have coprocessors for graphics and sound that are more sophisticated than most PCs sported even 5-6 years ago. I don’t think the limitation is so much lack of computing horsepower as it is lack of RAM, power consumption and wireless bandwidth. I’d also question how much raw data you really want to download and store on an easily-stolen portable device.
That last factor is what’s likely to drive analysis into the cloud over the next decade.