Users - entity sync

[davidnwelton]

> If this is what you are wanting to do, have a look at the entity sync
> stuff.

Oh, neat...something else interesting that I'd overlooked.

I found a little bit of info here:
OFBizBasicProductionSetup.pdf

"This uses the Entity Synchronization service in the Entity Engine and
the XML import file for this has pretty good comments about what does
what."

This phrase could probably use a reference to the file to make things
a bit clearer for those of us who aren't sure just where that file
might be.

Is the basic production setup guide something that could go into the
confluence server?  I have a pretty good grasp of how to get things up
and running on a Debian/Ubuntu system at this point, and wouldn't mind
adding that somewhere official.  I guess I should request a docs
account to try and contribute a few things back here and there.

Coming back to the sync stuff, it looks potentially very interesting,
because one of our needs, as we grow, is going to be to keep a local
copy of our data and system so that it's 1) faster, 2) takes load off
the main system, and 3) is still up in case our web connection is
down.  The (to me at least) obvious way to do that is to read locally
from a slave database, and send writes back to the main system.  What
role might the entitysync stuff play in that process?  I think if it
does anything with XML, the database route is going to be more
efficient, but would entitysync allow us to perform both reads and
writes locally and have that go back to the main system?

Thanks,
--
David N. Welton
 - http://www.dedasys.com/davidw/

Linux, Open Source Consulting
 - http://www.dedasys.com/
 
_______________________________________________
Users mailing list
[hidden email]
http://lists.ofbiz.org/mailman/listinfo/users

[Andrew Sykes]

David (Welton),

Sounds like your requirement is exactly what the sync stuff is set up to
do.

There is an example here...
applications/pos/data/PosSyncSettings.xml

In this case each POS terminal is syncing to a "Per Store Server" and
that syncs to a "Central Server".

Hope that helps.
--
Kind Regards
Andrew Sykes <[hidden email]>
Sykes Development Ltd
http://www.sykesdevelopment.com

 
_______________________________________________
Users mailing list
[hidden email]
http://lists.ofbiz.org/mailman/listinfo/users

[David E. Jones]

David Welton wrote:
> Is the basic production setup guide something that could go into the
> confluence server?  I have a pretty good grasp of how to get things up
> and running on a Debian/Ubuntu system at this point, and wouldn't mind
> adding that somewhere official.  I guess I should request a docs
> account to try and contribute a few things back here and there.

It isn't yet, but that's something on my todo list as Ruth Hoffman (who helped with the current document) has expressed interest in helping to update it.

I don't think something as specific as deployment on a certain flavor and version of Linux should go into the Basic Product Setup Guide, but it could certainly go in the general wiki. There may already be some of this sort of thing in the old wiki. You don't need any special permissions to create/edit pages in the new Confluence server (docs.ofbiz.org), just create an account and go to the wiki space and you can create pages right away.

> Coming back to the sync stuff, it looks potentially very interesting,
> because one of our needs, as we grow, is going to be to keep a local
> copy of our data and system so that it's 1) faster, 2) takes load off
> the main system, and 3) is still up in case our web connection is
> down.  The (to me at least) obvious way to do that is to read locally
> from a slave database, and send writes back to the main system.  What
> role might the entitysync stuff play in that process?  I think if it
> does anything with XML, the database route is going to be more
> efficient, but would entitysync allow us to perform both reads and
> writes locally and have that go back to the main system?

It doesn't intercept DB reads or writes, everything still runs locally against the database. All it does is asynchronously (and in the background) move data between servers as configured in the EntitySync setup.

-David
 
_______________________________________________
Users mailing list
[hidden email]
http://lists.ofbiz.org/mailman/listinfo/users

[davidnwelton]

> I don't think something as specific as deployment on a certain flavor and version of Linux should go into the Basic Product Setup Guide, but it could certainly go in the general wiki. There may already be some of this sort of thing in the old wiki. You don't need any special permissions to create/edit pages in the new Confluence server (docs.ofbiz.org), just create an account and go to the wiki space and you can create pages right away.

Makes sense.

> > Coming back to the sync stuff, it looks potentially very interesting,
> > because one of our needs, as we grow, is going to be to keep a local
> > copy of our data and system so that it's 1) faster, 2) takes load off
> > the main system, and 3) is still up in case our web connection is
> > down.  The (to me at least) obvious way to do that is to read locally
> > from a slave database, and send writes back to the main system.  What
> > role might the entitysync stuff play in that process?  I think if it
> > does anything with XML, the database route is going to be more
> > efficient, but would entitysync allow us to perform both reads and
> > writes locally and have that go back to the main system?

> It doesn't intercept DB reads or writes, everything still runs locally against the database. All it does is asynchronously (and in the background) move data between servers as configured in the EntitySync setup.

Does anyone out there use this?  I guess I'm trying to get an idea of
how safe it is.  I suppose the answer is that it isn't really be
default and that it requires thinking about the data being handled.
How would it handle something like two orders being placed on two
different machines?  Or something with a sequenceId?

--
David N. Welton
 - http://www.dedasys.com/davidw/

Linux, Open Source Consulting
 - http://www.dedasys.com/
 
_______________________________________________
Users mailing list
[hidden email]
http://lists.ofbiz.org/mailman/listinfo/users

[David E. Jones]

David Welton wrote:
> Does anyone out there use this?  I guess I'm trying to get an idea of
> how safe it is.  I suppose the answer is that it isn't really be
> default and that it requires thinking about the data being handled.
> How would it handle something like two orders being placed on two
> different machines?  Or something with a sequenceId?

Yes, ID sequencing needs to be pre-planned for data from multiple machines to coexist happily on a single target system. The Entity Engine can put a prefix on all sequenced IDs, configured on the delegator def in entityengine.xml. Of course, it depends on what the systems are for and the data that will be generated and tracked on each one. Usually just a few entities are moved over between the systems, and that should generally be as limited as possible.

Whatever the case, coordinating data between multiple systems is complex and what is being done should be well understood. It's not feasible to cover the potential problems and recommended solutions here, and of course those will change a lot depending on the circumstance.

-David

 
_______________________________________________
Users mailing list
[hidden email]
http://lists.ofbiz.org/mailman/listinfo/users