[Dev] ZODB for object persistence [Was: ZODB is not a Storage Technology]

[John Anderson]

Michael R. Bernstein wrote:

>On Fri, 2002-11-08 at 13:51, John Anderson wrote:
>  
>
>>Michael R. Bernstein wrote:
>>    
>>
>>>I *can't* advocate the ZODB much, because I'm no more certain of the
>>>requirements that led to it's selection than anyone else on this list.
>>> 
>>>      
>>>
>>We're interested in ZODB for a couple reasons: transparent object 
>>persistence is a big boon to productivity -- you don't have to write 
>>code to get data in and out of a database and dealing with data in 
>>native data structures is just much simpler.
>>    
>>
>
>Ok, that pretty much confirms my suppositions.
>
>  
>
>>As it turns out the Python 
>>community has figured this out and they are working on a proposal to get 
>>persistence built into Python 
>>(http://www.python.org/sigs/persistence-sig/). It looks to me like ZODB 
>>is a front runner for influencing the eventual Python standard for 
>>persistence.
>>    
>>
>
>Yes, this is correct, especially now that PythonLabs (Guido et al) are
>taking a direct hand in developing ZODB. In fact, I wouldn't be
>surprised if the ZODB *becomes* that standard.
>
>  
>
>>Trouble is, ZODB has a bunch of problems, for example it's 
>>dog slow. So I'd like to investigate what would be necessary to make 
>>ZODB a better solution.
>>    
>>
>
>Are you talking about write performance or read performance? And are we
>talking about the ZODB alone, or in the context of Zope?
>
My test program for object creation using ZODB4 goes from somewhere 
around 20,000 objects created per second to 300 per second once you 
throw in transactions. I'd like to use transactions for undo and 
resource exaustion recovery.

>
>Michael Bernstein.
>
>_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
>
>Open Source Applications Foundation "Dev" mailing list
>http://lists.osafoundation.org/mailman/listinfo/dev
>  
>