2

Further to this question about Blade vs Rack servers I wanted to check some stories I have heard.

Are Blade servers much noisier than regular rack servers? (or was that just when they first came out?)

also, how much more cooling equipment do they need?

NB: I'm probably talking about 1 chassis with 7 or 8 blades in it, compared to 7 or 8 normal rack servers. Not huge numbers of servers.

6

I would say anecdotally yes. They are more noisy. Although often more efficient in terms of overall power consumption, the act of pulling and pushing air through passive modular components with large front and rear facing blowers is always likely to create more noise.

Blades shouldn't need any external cooling equipment. Just a hot and cold aisle to deploy in ideally. The chassis takes care of the rest.

  • I am inclined to say no. Not noisier. More sense thus noisier in abssolute terms, but not noisier "per watt" or "per comutational unit". – TomTom Feb 13 '12 at 13:38
2

Blade servers are usually pretty silent - it's the fans in the enclosures that are noisy.

A single blade in an enclosure will be noisier than a single pizza-box server but a full blade enclosure should be quieter than the same number of pizza-box servers added up.

1

When we put in the HP Bladeserver rack back in 2005 the noise level in the datacenter increased markedly. What's more, the fans on that particular rack system had a particular whine harmonic that really drove me nuts. I brought my shooting earmuffs to work for when I need to be in there for any length of time.

That said, HP has seriously re-engineered their blade racks, so this may no longer be true.

1

Answer to a very old question -- but blade systems are dirt cheap on the secondary market now. We have a used HP C7000 Bladesystem chassis in our lab now, purchased for $200 fully populated with fans, power supplies, management modules, and a couple switches. It now has 3 blades, one storage array blade, two Cisco switch modules, and two 10Gb modules.

That package is far quieter than the screaming 1U Supermicro lab machines we had previously, and a lower pitch. As the lab is not in a data center, that change was very noticeable (and welcome).

When we previously ran two HP c7000 chassis full of blades in production in the timeframe of this question (2009), they were rather noisy of course, but that was in a data center. They seemed to comparable to the noise from a rack full of 2U servers and storage arrays, which use similar sized fans.

0

It depends.

I have some HP C3000 chassis, and when their fans are spooled up you can hear them through a door and down two flights of stairs. When the fans are throttled back, they are no noiser than any other relatively-powered pile of computers.

As far as cooling goes, a full chassis will use denser power-per-rack-unit, and will require correspondingly more cooling.

0

Most blades have smaller fans running at higher RPMs, to move the required amount of air, than a typical 2U+ rackmount server. Therefore, they tend to (all else equal) make more noise than rackmount servers (other than 1U boxen, which also have tiny fans).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.