[ZODB-Dev] BerkleyStorage

[Paul Everitt]

And if security doesn't make it a wash, DTML or ZPT will.  One important 
pattern here is to run BerkeleyStorage as the storage for your ZEO 
storage server, with perhaps FileStorage on all the clients.  Read time 
on this case is from the local caches (in memory or on disk, the latter 
as a file storage).

--Paul

(Apologies if this is posted as HTML.  Mozilla sometimes baffles me.)

Michel Pelletier wrote:

>Chris Withers wrote:
>
>>>My tests indicate that Full storage (i.e undoable, versioning) is
>>>about twice as slow as FileStorage.  Jim thought that that wasn't too
>>>bad.  :)  I haven't spent any time on performance profiling, but will
>>>likely do that after the 1.0 final release.
>>>
>>Probably a stupid question, but how come?
>>
>
>There's an entire extra layer of complex software between you and the
>database file.
> 
>
>>Does this affect reads from the DB itself or reads from the Zope object
>>cache as well?
>>
>
>I don't know about the first, but the object cache is above the storage
>object; its speed should not be effected.
>
>>In what ways would this performance decrease affect the overall performance
>>of Zope? (I suppose the main thing I care about is will there be a noticable
>>increase in the time Zope takes to deliver a page?)
>>
>
>Probably not, unless your can notice hundreths of milliseconds.  Really,
>the question is too broad.  There are many, many more circumstances that
>effect Zope's efficiency.  Compared to the amount of computation that
>Zope's security model must do, the extra time that BDB adds is probably
>a wash.
>
>-Michel
>
>_______________________________________________
>For more information about ZODB, see the ZODB Wiki:
>http://www.zope.org/Wikis/ZODB/
>
>ZODB-Dev mailing list  -  ZODB-Dev@zope.org
>http://lists.zope.org/mailman/listinfo/zodb-dev
>