Icon View Thread

The following is the text of the current message along with any replies.
Messages 1 to 2 of 2 total
Thread Replication Question
Fri, Nov 8 2013 1:52 PMPermanent Link

Uli Becker

I am storing quite large files in a published database and would like to
improve the performance of replication.

Table "Test" published on Computer A (Local Computer)
Table "Test" published on Computer R (Remote Server)
Table "Test" published on Computer B (Local Computer)

Computer A:
a) Insert a record (e.g. a large blob) "NewRecord"
b) Save updates
c) Send updates to R

Server R:
d) Load updates
e) Save updates

Computer B:
f) Download updates from R
g) Load updates

So far everything is normal.

Now:
How can I avoid that Computer B saves "NewRecord" again into an update
file? Since B is a client computer which sends changes only to the
remote server, that will result in sending the same (large) record back
to R. There it will be neglected.

I guess I can't avoid that, but it would save a lot of time and traffic
if there would be a kind of switch "DoNotPassReceivedUpdates". Or am I
missing something basic?

Thanks Uli
Wed, Nov 20 2013 2:37 PMPermanent Link

Tim Young [Elevate Software]

Elevate Software, Inc.

Avatar

Email timyoung@elevatesoft.com

Uli,

<< How can I avoid that Computer B saves "NewRecord" again into an update
file? Since B is a client computer which sends changes only to the remote
server, that will result in sending the same (large) record back to R. There
it will be neglected. >>

There's really no way to avoid this.  EDB can't possibly know which target
system will receive the updates, so the filtering on which update has been
"seen" already has to occur on the target system.  Therefore, the source
system must add all relevant updates into the updates file.

<< I guess I can't avoid that, but it would save a lot of time and traffic
if there would be a kind of switch "DoNotPassReceivedUpdates". Or am I
missing something basic? >>

I see what you're saying, but that is a very dangerous switch to have.  One
bad setting and you could mess up an entire database by excluding relevant
updates, which seems like a high price to save one trip from computer B to
the remote server.  After the updates reach the remote server again, they
will simply get ignored by the remote server if it has already seen them, so
they will "disappear" from the update files at that point.

Tim Young
Elevate Software
www.elevatesoft.com
Image