Microsoft Says Wireless Data Centers Possible

[TSGames]

A team of researchers from Microsoft and Cornell University has concluded that, in some cases, a totally wireless data center makes logistical sense.
In a new paper, a team of researchers from Cornell and Microsoft concluded that a data-center operator could replace hundreds of feet of cable with 60-GHz wireless connections—assuming that the servers themselves are redesigned in cylindrical racks, shaped like prisms, with blade servers addressing both intra- and inter-rack connections.

http://slashdot.org/topic/datacenter...nter-possible/

----------------------------------------------
Interesting use of the newly approved WiGig technology. The same technology that will soon be infiltrating a living room near you.

It is interesting.....I'd just be concerned with what happens when a tech walks in front of one of the servers....at 60 GHz it doesn't have a whole hell of a lot of penetration, and in a data center one body could be taking out gigabits of information per second. Feasbile, but I'm not sure if it's practical just yet.

Any thoughts?

[Mando Knight]

If they're tri-band, I think they'd be using the 60 GHz for a phased-array or other directed radio wave system rather than a broadcast, so they can boost the power more easily and/or guarantee that they have line of sight to the receiving end.

That, and the other two bands could serve as a link to ensure that the systems know when their signals are being interrupted and are able to compensate.

[Anarion]

How likely would it be to cause cancer to anyone working in the place?

[Weezer]

How likely would it be to cause cancer to anyone working in the place?

Unlikely. GHZ radiosignals are not even close to high enough energy for that. Light must be 'ionizing' to cause cancer, and that happens only at high energy, it starts somewhere around 200 nm, in the extreme UV. This converts to about 1499 Terahertz, or ~1500 times as much energy as the 60 GHz they're talking about.

[TSGames]

If they're tri-band, I think they'd be using the 60 GHz for a phased-array or other directed radio wave system rather than a broadcast, so they can boost the power more easily and/or guarantee that they have line of sight to the receiving end.

That, and the other two bands could serve as a link to ensure that the systems know when their signals are being interrupted and are able to compensate.

tri-band like a cellphone...?

[Mando Knight]

tri-band like a cellphone...?

More or less. The WiGig specifications call for 2.4, 5, and 60 GHz bands, rather than the lower frequency bands used for cell phone communications. 2.4 is the current frequency for Wi-Fi systems, so it has approximately the same range, data transfer speed, and penetration that you've probably become accustomed to. I imagine the 5 GHz band will be used when 60 GHz isn't feasible or is just plain overkill (it takes more power to run a signal at a higher frequency), but 2.4 GHz isn't enough.

[Maelstrom]

Interesting idea, but they need to work on the speed a bit...10 GB is already being pushed to the limits on virtualized stacks, and with 40G networks on the near horizon the 4-15Gbps the article states may be a bit behind the curve, however interesting the design looks...

[Archonic Energy]

Stop it. You are getting my work in the playground.
I come to hide from work...

Ok. Having read the paper. Not a bad idea, you`ll need to sheild the dc a bit to lower the rf noise between suites and to decrease the chance of a wardriver.

I wonder when CISCO will go wireless on their 6500 series...
Really should speak to Z...

[grimbold]

wow this is... really cool
thanks for sharing!

[TSGames]

More or less. The WiGig specifications call for 2.4, 5, and 60 GHz bands, rather than the lower frequency bands used for cell phone communications. 2.4 is the current frequency for Wi-Fi systems, so it has approximately the same range, data transfer speed, and penetration that you've probably become accustomed to. I imagine the 5 GHz band will be used when 60 GHz isn't feasible or is just plain overkill (it takes more power to run a signal at a higher frequency), but 2.4 GHz isn't enough.

I see what you mean now.

I did some looking into it: the transceivers used in the paper only support the 60 GHz frequency. This is most likely due to the increased space, heat, and energy footprint of having transceivers for 2.4 and 5 Ghz.

Then I got curious about what was meant by "Tri-band" from this line of the wiki:

WiGig tri-band enabled devices, which operate in the 2.4, 5 and 60 GHz bands, will deliver data transfer rates up to 7 Gbit/s, about as fast as an 8 antenna 802.11ac transmission, and ten times faster than the highest 802.11n rate, while maintaining compatibility with existing Wi-Fi devices.

After a little bit of research it became apparent that the wiki is either intentionally misleading or poorly written. A "Tri-band enabled device" is really just a "Backwards compatible device with respect to Wifi on the 2.4 and 5 GHz frequencies". In short, Tri-band is not necessary to achieve maximum throughput and has no impact on the throughput at all; it's merely a marketing label for devices that have built in Wifi frequency hardware.

The original article didn't go into enough detail, but I think the inward facing transceivers may be able to help compensate for signal interruption by distributing lost data to other servers in the cluster.

Also, I'm not sure that additional shielding is necessary; although additional protection is always good, it does add a bit of cost. It seems like putting the servers in rooms that don't have windows would probably keep them safe from wardriving. The 60 GHz signal doesn't really deal with walls too well, so anyone listening in is gonna have to explain to security why they are pulling a Splinter Cell. Of course it would still be open to creative methods of tapping....but then again, so are wired data centers.

[Archonic Energy]

I'll admit the Sheilding I suggested was to alleviate the customers fears of wifi... it has a rep for being unsecure, so if you were going to sell it you'd need to counter that fear somehow, grounded aluminium foil in the wall cavities should be sufficent.

[TSGames]

That's a good point. Now you have me curious about cheap shielding methods... TO BING AND AWAY! lol

[SiuiS]

How likely would it be to cause cancer to anyone working in the place?

It terrifies and amuses me that we both had the same foremost thought.