April, 2009

IBM paying Sun users to abandon SPARC following rejection

Maybe Sun Microsystems shouldn’t have said no to IBM’s acquisition offer, because it appears Big Blue doesn’t like rejection.

IBM doubled the dollar amount for its Power Rewards migration services specifically for Sun customers that switch from Sun SPARC, UltraSPARC or SPARC64 processor-based servers to IBM Power Systems.

Perhaps IBM is just taking the opportunity to catch Sun defectors from the upcoming Oracle takeover, but when IBM talked about acquiring Sun, there was speculation that the company would kill SPARC systems, which compete with IBM’s Power Systems.

So now that IBM can’t outright squash Sun’s SPARC system, the company hopes to get rid of SPARC another way; by giving Sun users an offer they can’t refuse.

IBM increased the money it offers to switch from competitive systems to IBM through its Power Rewards Program from $4,000 to $8,000, based on the number of SPARC, UltraSPARC or SPARC64 microprocessors used in each Sun or Fujitsu server.

Customers can use the migration services to move workloads running on Sun SPARC-based servers to IBM’s AIX, Linux or i operating environments.

For example, a Sun customer using a Sun Fire V890 system with eight microprocessors (or “cores”) would now receive $64,000 in migration services, up from $32,000, to move the workloads to an IBM Power system.

Since Oracle has not addressed their plans (or returned calls and emails to answer questions) about the fate of Sun’s hardware and SPARC systems under Oracle that this incentive will probably work quite well.

Boston Marathon website prepares for traffic surge

About a million people will visit the Boston Marathon website on Monday to check out the 113th annual race, so the IT pros supporting the website have been hard at work these past few weeks making sure the site doesn’t crash that day.

The Boston Athletic Association (BAA)’s technical director, John Burgholzer, spends three weeks prior to the race building the infrastructure at a colocation facility in Massachusetts to support the BAA’s website and other technologies surrounding the marathon, like the new AT&T Athlete Alert System, which delivers text messages to people who are tracking runners whenever their runners hits a checkpoint.

Burgholzer, who owns a technology consultancy company in North Reading, MA called Information Overload, uses Hewlett Packard (HP) blade servers to run everything. Interestingly enough, he doesn’t go the virtualization route, which would probably be quicker and easier, because he doesn’t know enough about it or trust the technology to handle the surge of users during race time.

“We haven’t tried out virtualization at all and I’m not sure we would. We get about 50,000 to 60,000 concurrent connections at peak time during the race, and I’m not sure virtualization would work for us performance wise,” Burgholzer said.

I’ve heard this apprehension about virtualization before, so it appears the technology is not as pervasive as companies like VMware would have us believe. Mainly because guys like Burgholzer are far too busy to learn about an entirely new technology, especially when their traditional approach works just fine for them.

So, Burgholzer adds seven HP ProLiant blade servers to the two that are typically used to run the website, for a total of nine blades running Windows 2003. HP blade servers were the right choice for the BAA because the servers require little space and are easier to manage that rack mount systems, plus, the organization had been using HP gear even before Burgholzer came on board nine years ago, and HP has always been “extremely helpful” at race time, he said.

Before the BAA moved from “a bunch of pizza boxes” to HP c-class blade servers in 2007, cabling and management “was a nightmare,” Burgholzer said. “We would build the data center up before the race using rented systems and people didn’t really care how it was set up, so we had a rats’ nest of cables in the back of the rack,” he said.

By switching to blade servers, the cabling is not an issue; he just slides new blades into the chassis as needed, and the management software makes configuartion easy, he said. The chassis has one gigabyte Ethernet connections on both the front and back ends of the server chassis, which he says are plenty, and he uses F5 Networks technology for load balancing.

With all of that, he’s confident there won’t be any issues with the website on Monday - knock on wood. “We have a pretty well-tuned website now; there was a bit of a bandwidth problem in 2007 and at the peak of the race we have seen the website running slower, but we have gotten it down now,” he said.

The day after the race, Burgholzer will start looking for ways to improve the website and new features to add for next year.

Google’s server recipe no longer secret

Looks like Google has finally spilled the beans on its server design, which the company has kept secret for years.

According to a report, Google let everyone see its design at a conference this week. The system was a 3.5 inch thick 2U system with two processors, two hard drives, and eight memory slots mounted on a motherboard built by Gigabyte. Google uses x86 processors from AMD and Intel, and the servers are powered by 12-volt batteries in case there’s a problem with the main source of electricity, according to the report.

We already knew that Google uses 12-volt power supplies for its servers, which the company says is more than 90% efficient compared with typical server efficiencies at or below 70%.

For the most part, Google runs these servers not out of brick and mortar data centers, but from those shipping containers that have become all the rage with vendors like Sun Microsystems , IBM, HP and Rackable Systems in recent years. Google was putting data centers in mobile containers long before those guys, and typically uses standard 1AAA shipping containers to house around 1,160 servers each, according to the report.