It would help to know what your key is.  If you only select by country and 
region, then having those as the first part of your key would speed up your 
retrieval.  

In general, if you want fast look up in hbase, the thing you want to use to do 
the lookup has to be in the key.  This sometimes leads to having more than one 
table holding similar information (sort of like how an index works under the 
covers).

Dave

-----Original Message-----
From: Laurent Hatier [mailto:[email protected]] 
Sent: Thursday, July 28, 2011 6:49 AM
To: [email protected]
Subject: GeoIP with HBase

Hi all,

I use HBase for the GeoIP operation.
I have a column family "GeoIP" where is 4'000'000 million lines.
Select a country&region from this column family takes me 7-8 seconds and
when i put this table IN_MEMORY, it's take about 5-6 seconds.
But, I must calculate the country&region for each log line that I stored
into HBase and the problem is I have million and million lines to treat...

So there is an another solution to calculate more quickly ? I've heard that
MongoDB is faster to do a select like this...

Thanks

-- 
Laurent HATIER
Étudiant en 2e année du Cycle Ingénieur à l'EISTI

Reply via email to