db-derby-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mike Matrigali <mikem_...@sbcglobal.net>
Subject Re: out of memory when writing blobs
Date Wed, 14 Mar 2007 17:27:39 GMT


Øystein Grøvlen wrote:
> For tables with Blob or Clob columns, the default page size for the will 
> be 32 kB.
> (Ref. http://db.apache.org/derby/docs/dev/tuning/ctunperf816059.html )

Currently Derby will never have 2 different page sizes in the same table 
or index.  As Oystein points out Derby will automatically choose 32k 
page size if it determines at create table time that is appropriate. 
Users can override this choice.

So blobs tend to make page cache bigger because page size is bigger,
nothing really to do with overflow pages.

> 
> -- 
> Øystein
> 
> Anders Morken wrote:
> 
>> Anders Morken:
>>
>>>  [...] the page size (Seemed to be 32K in this case [...]
>>
>>
>> Aha. Those 32 kilobyte pages were "overflow pages", created to hold
>> rows that won't fit on a regular page. The regular pages were 4
>> kilobytes, the default size, but obviously those 6 megabyte blobs had
>> to be split across a few overflow pages. =)
>>
>> That explains why blob workloads suddenly consume a lot more memory than
>> regular workloads with few (if any) overflow pages. Maybe this is
>> common knowledge, but I just hadn't realized. =)
>>
>> Sorry for the noise, just wanted to clear up (my own) confusion. =)
>>
> 
> 
> 


Mime
View raw message