huang zuxing blog @ home the quieter you become, the more you are able to hear.

23Nov/14Off

HP Apollo 6000 (2014)

After deployed HP Apollo 6000, like to high light below limitation as HP sale team  didn't mention those before :

1.  Putting 4 or 5 switches in the middle of rack is not a good idea,  hot air come out from the middle of rack and suck into top level chassis, hot spot creates ;  recommend to put switch on top of rack: [ updated this on 12/Dec/2014, at the end after read HP doc by ourselves, we found HP network team sent us the wrong fan model, which push hot air to front cold aisle although we have set fan direction " power to port" , after replaced the fan with correct model, no more such issue.]

17186-019a-Door-off-lo-res_apollo6000 IMAG3180_apollo_heatp

2.  APM reports node power wrongly, in the max load,  it reports every compute node uses 170w+ ; total 140 nodes  x 170w  = 23.8kw , but the actual is,  total whole rack , 140 nodes  use power 13kw only .

Apollo_APM_power_report1 Apollo_APM_total_power_report

 

3.  Power shelf limitation:  for PSU,  it is  N+N;  but if the AC to DC board is failed, all chassis which use this power shelf  will be power off,  each power shelf could connect 6 chassis, the impact is huge  .

001zu03_apollo_power

 

4.  Each Chassis has 5  fan modules, each  module has 2 fans; if one of module is down, whole chassis will be down.  if one of 10 fans is down,  to replace the fan module, engineer must put in new module within 60 seconds after remove the faulty fan module.  if not , the whole chassis might be down.

 

18Nov/14Off

Rear Door Heat Exchange from coolcentric Nov/2014

IMAG3190_DCIMAG2828_DC

18Nov/14Off

tongtong on 11.10

我可爱的小三。

IMAG3091IMAG3085

IMAG3151IMAG3153-MOTION

IMAG2952

18Nov/14Off

the simplest way to explain mapreduce

Below copy from internet:

下面这段话是网上其他人用最简短的语言解释MapReduce:

We want to count all the books in the library. You count up shelf #1, I count up shelf #2. That's map. The more people we get, the faster it goes.

我们要数图书馆中的所有书。你数1号书架,我数2号书架。这就是“Map”。我们人越多,数书就更快。

Now we get together and add our individual counts. That's reduce. 

现在我们到一起,把所有人的统计数加在一起。这就是“Reduce”。 

Filed under: IT Comments Off