[Monetdb-developers] SQL concurrency/serialization

Martin Kersten Martin.Kersten at cwi.nl
Thu Feb 25 19:53:26 CET 2010


Isidor Zeuner wrote:
> Dear MonetDB developers,
Dear Isidor,

MonetDB/SQL follows an optimistic concurrency control scheme.
One of the properties of OCC is that it does not lock the
database, except for validation of potential conflicts between
the r/w sets.
As such, it favors low update frequency.

There are no facilities to control, take control over this behavior.

If you aim for a (extreme) high update rates you could
consider a master/slave setup, where the master executes
the update in serial way. In most cases this is more efficient
as well.

Alternatively, a middleware layer can perform a coarse grain
locking scheme.

regards, Martin

> 
> on concurrently running writing transactions, I am getting a rather
> narrowly descriptive error message: "!SQLException:sql.trans:COMMIT:
> failed".
> 
> My first attempts to deal with this issue suggest that it happens
> exactly in those cases where the concurrent writes create some sort of
> conflict.
> 
> I am aware that MonetDB does not focus on concurrency, and of the
> suggestion to serialize multiple writers. However, I'd like to
> know what the current status of concurrency is. To what extent are SQL
> primitives implemented which allow to let the server take care of
> serializing writes, for example through explicit table locking or
> transaction isolation?
> 
> Best regards,
> 
> Isidor Zeuner
> 
> ------------------------------------------------------------------------------
> Download Intel® Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> Monetdb-developers mailing list
> Monetdb-developers at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/monetdb-developers





More information about the developers-list mailing list