[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: LedgerSMB Scalability Bottleneck-- proposal



Hi;

On Dec 12, 2007 8:45 PM, Joshua D. Drake <..hidden..> wrote:
Chris Travers wrote:


As I understand it, a timeout happens because nothing is happening. E.g;
 there is no communication between the apache server and the client.
Why not just have the stored procedures spit out an update as they are
working?

The big issue is that this does not play well with the template system.  We can't actually have another round trip without either committing the batch save part-way through or reworking our template system.


Processed 100 transactions
          200 transactions
          300 transactions

If we go that way, we can't readily use templates because they send out full HTTP headers,  I suppose we could simply bypass the templates.  Or issue commits part way through and add extra roundtrips to the web server.  Neither of these sound good to me though :-(  The former breaks maintainability of the app, and the latter breaks maintainability of the data....

The other issue with this specific class of users may end up eventually running into timeout issues in the template processing for these sorts of payment processing.  A queue system might provide a solution for that as well.


Until done. If you wanted to offset load, we can use plperl to implement
some sleep patterns.

At the moment I don't see any reason for sleep patterns.  My current queue system doesn't use any sleep patterns.


Joshua D. Drake


-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
Ledger-smb-users mailing list
..hidden..
https://lists.sourceforge.net/lists/listinfo/ledger-smb-users