Google Reveals "Secret" Server Designs

Hugh Pickens writes "Most companies buy servers from the likes of Dell, Hewlett-Packard, IBM or Sun Microsystems, but Google, which has hundreds of thousands of servers and considers running them part of its core expertise, designs and builds its own. For the first time, Google revealed the hardware at the core of its Internet might at a conference this week about data center efficiency. Google's big surprise: each server has its own 12-volt battery to supply power if there's a problem with the main source of electricity. 'This is much cheaper than huge centralized UPS,' says Google server designer Ben Jai. 'Therefore no wasted capacity.' Efficiency is a major financial factor. Large UPSs can reach 92 to 95 percent efficiency, meaning that a large amount of power is squandered. The server-mounted batteries do better, Jai said: 'We were able to measure our actual usage to greater than 99.9 percent efficiency.' Google has patents on the built-in battery design, 'but I think we'd be willing to license them to vendors,' says Urs Hoelzle, Google's vice president of operations. Google has an obsessive focus on energy efficiency. 'Early on, there was an emphasis on the dollar per (search) query,' says Hoelzle. 'We were forced to focus. Revenue per query is very low.'"

Comments

Hey google, want to save some money?

[geekoid]

Get some mainframes.
For cryinh out load, with 1 mainfram you can't have a mainframe with 30,000 or more intances of your operating enviroment on it. Possible up to 100K.

Put 5 of these in each data center. Cheaper to power, you would only need a few people to keep it running, it would run for 20 years.

Want to save more money, here is another way:
Build your data center in the desert and build 150 MW industrial solar thermal system to power it. Sell the extra power.

Oh, and if you are not up to date on Big Iron, don't fucking reply becasue your going to look like a fool.

Re:Hey google, want to save some money?

[sexconker]

There is no possible way their solution is cheaper than a real mainframe (created for the task) when all costs are considered.

Nor is there any possible way their solution is more reliable, or more "green".

Didn't we all learn about how Google buys the cheapest, crappiest hard drives it can and just RAIDs them and tosses them out when they break?

This is the same thing they're doing here with "commodity" hardware. They buy the cheapest, oldest, used shit that runs and wire it all together.

The real issue here is not the numbers, but Google's desire to be completely independent from other vendors. It's why they "build" their own storage arrays and switches.

If they can get some PR about how they're so cool and green while they're at it, they'll do so, despite it being bullshit.

Just look at the pictured server:

Dual socket Gigabyte board (GA-9IVDP) (fine) with electrolytic caps (not fine, NOT server-grade).
Xeon Noconas, I think. Only 5 years old.
Aluminum heatsinks.
DDR2 400.
Trash (Magnatek) power supply.
Hitachi Deathstars.
A 12v battery. I never knew DC was more efficient than AC! WOW GOOGLE IS SO COOL!

A good mainframe would last decades. Google's frankenframe (lets call it what it is) must be sloughing off parts like skin cells from a Texan with eczema.

Re:The New Mainframe

[AKAImBatman]

Dude, seriously. Would it have hurt you to read the FIRST *BLEEPING* REPLY POSTED OVER AN HOUR BEFORE YOURS AND MODDED UP TO +4 FUNNY??? You know, the one where I complained my brain was going at my old age, and that I meant to type "microcomputers"?

Maybe? You think? Possibly? Perhaps? Just slightly? Bueller? Bueller? What was I saying again? This is a lot of question marks? If you can read this bumper sticker, you are too close?

Here in real world, mainframes continue chugging along handling the tasks that mini's, workstations, and PC's can't.

Here in the real world, traditional mainframes are maintained for legacy support. The other class of mainframes (e.g. the Unix-based ones) are quickly falling out of favor for new, large-scale projects. Making scalable clusters of computers is a workable solution that provides a greater deal of flexibility than plopping down 5-10 big iron machines.

Not that Big Iron is going anywhere. It's just become the modern minicomputer compared to these new container-mainframes.