ws-sandesha-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jaliya Ekanayake" <jal...@opensource.lk>
Subject Re: [Sandesha2] Changes to the permanent storage model
Date Thu, 20 Oct 2005 16:35:14 GMT
Hi Chamikara,

Can you please explain whether this is problem for transaction or for RM. I mean we can either
use Axis2's database cofiguration or our own database (I prefer the latter since then we have
more control over what we are doing - we may need to give a configuration scripts for different
databases)

Can you pleaes explain this a bit more
>This is because we cannot maitain a transaction that combine several store operations.
So if something goes wrong in the middle of two storage >operations we won't be able to
do much about this when the system restart.

Thanks,

Jaliya
  ----- Original Message ----- 
  From: Chamikara Jayalath 
  To: sandesha-dev@ws.apache.org 
  Sent: Thursday, October 20, 2005 11:15 AM
  Subject: [Sandesha2] Changes to the permanent storage model


  Hi All,

  Current model of Sandesha2 aims to fascilitate true reliability (persistance based) by using
the Context hierarchy serialization fascility provided by Axis2. Sandesh2 saves its tables
in the configuration context. 

  But, as Paul pointed out today, there is a defect in this design. This is because we cannot
maitain a transaction that combine several store operations. So if something goes wrong in
the middle of two storage operations we won't be able to do much about this when the system
restart.

  So it seems that we will have to move to a database based model (architecture of the current
implementation fascilitates this). We will have to design transactions carefully to make sure
that the related sets of storage operations are done as a atomic unit.

  Thoughts ???


  Chamikara 
Mime
View raw message