Month: June 2009

Antikythera

by

Yesterday at the Usenix Technical Conference in San Diego, I listened to a delightful talk about the Antikythera Mechanism, a 2000-year-old bronze object recovered from a shipwreck in 1901 but only later recognized as an astronomical calculator.

The speaker, Diomidis Spinellis, has written a software emulation of the mechanism, allowing the viewer to see how the different sets of gears move. There’s a particularly ingenous pair of gears known as k1 and k2, which help represent the Moon’s elliptical orbit.

Robots in space

by

Today the LCROSS mission launched from Cape Canaveral in Florida. On October 8th it will intentionally smash into the Moon’s south pole. The resulting debris cloud will be analyzed for traces of water, as hidden valleys at the Moon’s poles are thought to contain large amounts of water. If so, then these could be used by future lunar exploration stations. Thus LCROSS is very much a preparatory mission for manned missions to the Moon.

To which I say “heart in the right place, but misplaced science priorities.” We need to admit here that the reason for space exploration is science. We must also admit that, in space, robots are better at science than we are. Human space travel is a bad idea for many reasons:

  • Space is dangerous. It’s full of radiation from the sun, including cosmic rays. There is no realistic way to shield them. Long-term space exposure will eventually cause death by radiation.
  • Space is cramped. Unbelievably claustrophobic. We have been acclimatized to the bright and cheery bridge of the Starship Enterprise, but the reality is about two cubic meters per person. Have you ever seen the Apollo 11 capsule at the Smithsonian in Washington? Six months to Mars and back in a container this tiny? This way lies madness.
  • A manned mission requires a return journey. This requires double the fuel & resources.
  • Politically, dead astronauts are very bad. What happens during an 18-month manned mission when an astronaut develops appendicitis or an aggressive kidney infection?
  • Manned missions cost several orders of magnitude more than a robotic mission. The extra systems and personnel required to keep fragile humans alive are daunting.

In contrast, robotic missions are a piece of cake. While not cheap in absolute terms, they cost a fraction of a manned mission (even a manned mission to the International Space Station). Recent missions have been fabulously successful: Cassini, Mars Rovers, Galileo, Ulysses, Deep Impact. Several other missions are en route to their destinations with all systems go: Messenger, New Horizons, Dawn.

Robots rule!

Cloudy Weather – Is My World Changing?

by

I’m a systems administrator. When asked what that means, I’ll say different things to different people. Sometimes I say “I keep the computers working,” other times I reply “I spend a lot of my time down in the data center.” I had the pleasure of designing my company’s new data center when we moved to San Ramon in 2006. And while a data center is a noisy, cold, windowless room, at least it was my noisy, cold, windowless room – and it was a damn sight better than the old noisy, cold, windowless room.

But will cloud computing change my world? If companies gradually abandon their local data centers and move everything to the cloud, there isn’t going to be any noisy, cold, windowless room. What then happens to their sysadmins, who were that glue between users, applications and hardware? Well, it comes down to whether you think the Cloud is the Borg, or just a bunch of marketing hype.

In one corner are bloggers like Mark Mayo, who predict Cloud Domination and suggest how sysadmins can prepare. Arguing for the defense, m0j0 believes the Cloud is hype overload.

I’m currently attending the Usenix Technical Conference in San Diego, and have been asking many of my colleagues what they think. Most are in the “hype” camp: interesting technology, but they are not going to take over the world. Stay tuned.

Clouds

by

This past Thursday I attended Fujitsu’s third annual Cloud Computing Symposium in Sunnyvale. It was an excellent program, with a succession of interesting panel discussions.

With cloud computing, you exchange applications and data on your own machine for applications and data somewhere out on the Internet. Gmail, for example, is a cloud-based application: all the data is up on the Net, and you can access it from anywhere.

The Cloud is hot. The Cloud is the new black. The Cloud is where you should invest. The Cloud will spark the next bull market in technology. Well, maybe. There’s certainly a lot of momentum and media attention – not to mention money – behind the Cloud.

And at first glance, you’d be crazy not to use the cloud. Your data is available from any Internet connection, be it in San Francisco or Timbuktu. No more worrying about upgrading applications, replacing a crashed disk, or backing up your data – all those niggling things that you’re not very good at. No more having to lug around a traditional laptop with a large disk. Heck, you can get to the cloud from your iPhone.

The economic advantages are even greater for businesses. Running a data center is a huge capital expense – construction, power, HVAC, all the servers, network hardware & cabling – as much as 40-70% of non-personnel expenditures. For a startup company these expenses can be overwhelming, and it’s all up front, before any revenue comes in the door. But if you use the Cloud, you simply rent your computers & applications (using operating expenses instead of capital expenses) and you’re ready to go almost immediately.

Viewed from a security standpoint, however, you’d be crazy to use the cloud. You are trusting that your vendor will keep your data private, both from themselves and from third parties. You are trusting that they make backups of your data. You are trusting that they have the wherewithal to maintain high uptimes. On Wednesday 10 June a lightning strike damaged an Amazon data center and many customers were offline for six hours. Vendors aren’t necessarily liable for these outages: many cloud vendors do not offer guarantees, or ask you to pay extra for “geographic redundancy.”

When a potential customer is leery of the cloud because of security concerns, a vendor will typically reply “are the data any more secure at your own site? The largest threat to any organization is from insiders.” I’m willing to acknowledge that, but moving to the cloud can’t be just that simple, or everyone would already be doing it. It’s a complex process.

There are legal ramifications of the cloud as well: whereas data in your possession can only be obtained by a search warrant, data in the cloud need only be subpoenaed – and sometimes the Cloud Powers that Be simply hand over data to the government upon request. And if the cloud is physically located outside the country, all bets are off. How do you know whether someone on the cloud is looking at your data? What if your trade secrets are leaked? Encryption is certainly part of the ultimate solution, perhaps in a way where data are only visible unencrypted at the customer’s site. This is an area of much research.

Despite the hype, cloud computing is still a small percentage of the industry, about 2% of the U.S. infrastructure according to one estimate. But in two or three years, that number will undoubtedly be much higher.

Welcome to the Cloud.