hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dru Jensen <drujen...@gmail.com>
Subject Re: correct performance evaluation results?
Date Tue, 28 Oct 2008 22:22:05 GMT
Did your scan have a column specified that doesn't exist?

On Oct 28, 2008, at 3:10 PM, Krzysztof Szlapinski wrote:

> stack pisze:
>> Yeah.  Unless you got some magic going on in that Xeon of yours.
> No magic noticed ;)
> But 4 real - Any ideas why the scan test goes wrong? I got no  
> warnings, no errors, nothing suspicious in the log files. It starts  
> and ends within aprox. 40ms.
>
> krzysiek
>
>> St.Ack
>>
>>
>> Krzysztof Szlapinski wrote:
>>> stack pisze:
>>>> Your math looks wrong.
>>>> St.Ack
>>>>
>>> Hm, I double checked and it seems OK
>>> Maybe the confuson is caused by the use of wrong decimal sparator  
>>> in my results (the Polish way ;) ). So once again its:
>>>                               no of rows      miliseconds     rows/ 
>>> sec
>>> sequential writes        2097140            408641     ~5132
>>> scan                          2097140                    40      
>>> ~52428500
>>> random read              4194280            843962    ~4970
>>>
>>> scan results seems tooooooo good ;)
>>>
>>>>
>>>> Krzysztof Szlapinski wrote:
>>>>> hi
>>>>> I've run the perormance evaluation test on a two machine cluster
>>>>>
>>>>> hardware:
>>>>> dual Xeon E5430 2.66 Ghz, 4GB RAM,
>>>>> software:
>>>>> OS - 64bit Ubuntu 8.04 Server
>>>>> Java 1.6 - 64bit
>>>>> Hadoop - 0.18.1
>>>>> HBase - 0.18.0
>>>>>
>>>>> Below are the results - the scan test result seems not real -  
>>>>> could someone please confirm that such result is credible
>>>>>
>>>>>
>>>>>    no of rows      miliseconds     rows/sec
>>>>> sequential writes     2097140     408641     5 131,99
>>>>> scan     2097140     40     52 428 500,00
>>>>> random read     4194280     843962     4 969,75
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> krzysiek
>>>>>
>>>>
>>>>
>>>
>>
>>
>


Mime
View raw message