May 2010 Archives

Lance Albertson wrote up a great description of how Ganeti Virtualization Manager performed under pressure during a power outage:

Nothing like a power outage gone wrong to test a new virtualization cluster. Last night we lost power in most of Corvallis and our UPS & Generator functioned properly in the machine room. However we had an unfortunate sequence of issues that caused some of our machines to go down, including all four of our ganeti nodes hosting 62 virtual machines went down hard. If this had happened with our old xen cluster with iSCSI, it would have taken us over an hour to get the infrastructure back in a normal state by manually restarting each VM.

But when I checked the ganeti cluster shortly after the outage, I noticed that all four nodes rebooted without any issues and the master node was already rebooting virtual machines automatically and fixing all of the DRBD block devices.

Ganeti is a management layer that makes it easy to set up large clusters of Xen or KVM (or other) virutalized machines. He has written a great explanation of what is Ganeti and its benefits too.

I use Ganeti for tons of projects at work.

Standford's team (which won the DARPA contest) is doing some great stuff.

When jet engines were new jet airplanes had to have 3 pilots. The third pilot did nothing but run the jet engines: constantly adjusting the settings, tuning them, and keeping them running manually. Eventually electronics were developed to the point that such controls could be automated, thus eliminating the third pilot. The electronic control system is not just less expensive, but it produces better fuel efficiency.

Cars today require a driver. The driver requires good health, must be rested and sober. Humans are not very good at optimizing fuel efficiency. Humans don't communicate very well between cars. Imagine a world where cars drove themselves. The computers could optimize for better fuel performance, people could relax during their commute, and the cars could network to get better performance. For example, if 10 cars were all driving to the same destination they could get into a line and drive like a 'train' eliminating wind resistance for each other. Who knows what other optimizations will be discovered: the lead car could take on different computational responsibilities than the other cars.

One problem with our current highway system is that we equate "safety" with "speed". What we want is a "safety limit" but that is hard to quantify so we make due with a reasonable approximation: the speed limit. Computer controlled cars could enable a true safety limit and be permitted to drive at any speed as long as their metric is maintained (super fast on straight roads, slowing down for curved roads or during rain). Wouldn't you prefer a driver that had a mathematical model of friction ratios based off of sensors on the tires?

Of course, as the "driver" we humans could select from a wide menu of maneuvers that are humanly impossible. Like, parking a car James Bond style.

Eventually the cost and safety issues will be worked out. At that point, autonomous cars may be a big time management win. In the meanwhile, the bar association should advocate for more research in this area. I don't mean the legal organization, I mean the association of bar owners!