Login ProductsSalesSupportDownloadsAbout |
Home » Technical Support » ElevateDB Technical Support » Support Forums » ElevateDB General » View Thread |
Messages 11 to 16 of 16 total |
Database Design |
Mon, Oct 19 2009 7:37 AM | Permanent Link |
Tim Young [Elevate Software] Elevate Software, Inc. timyoung@elevatesoft.com | Tony,
<< Is it the case that when replicating updates from a main database to a group of client databases, each client will receive an update file containing all updates for all clients? And that each client will need to load this all-inclusive update file into its database before somehow removing the updates that are not relevant to the client? >> Correct. You can also just use BEFORE triggers to raise exceptions that are then eaten by ERROR triggers, but that's a little convoluted. << If I've understood this correctly, then the ability to create individual update files at the master database which are specifically tailored for each client would be very handy. >> Yes, the only issue is the current storage mechanism. -- Tim Young Elevate Software www.elevatesoft.com |
Mon, Oct 19 2009 7:43 AM | Permanent Link |
Tim Young [Elevate Software] Elevate Software, Inc. timyoung@elevatesoft.com | Leslie,
<< I suppose the actual writing to the change log file is triggered by committing a transaction. Is it possible to apply a SQL based filter during that procedure for the tables? If yes, we could have a set of filters with different change log files as target. So the PUBLISH command could have a an extra INTO STORE parameter and the every table could have a FilterCondition parameter. This would mean we have more than one publish rule set so every PUBLISHING should probably have a unique name for intended the recipients too. >> The store functionality probably won't work since right now the published updates are logged in the database directory along with the normal table files, but you do present an interesting solution. There might be a way to use separately-named update storage files for each, thus avoiding any issues with fragmentation of the files. << If this is possible it could be a very flexible solution, but with linearly growing performance cost. The more log files there are, the slower committing a transaction is. So it would probably be better to save a complete change log first and process it in a lower priority thread where or the other replication targets could be created during a simulated LOAD UPDATES. >> Possibly, although the update logging is usually quite small, so I'm not particularly concerned with commit times. << If RecepientGroupName comes into the picture as a new system object, it might be a logical extension to have a detail system table with the URL of the databases which should receive the updates. So at the and we could have an easily configurable target list plus a new command like >> Updates files (the standalone .edbupd files, not the files used for storing the published updates) are stored in stores, and transferred to and from locations, by default, using local and remote stores. So, there really isn't any concept of a database with respect to these files. However, one can easily copy any given update file to numerous stores quite easily, and one can even define a table in the database for storing information about which stores get which update files. Scheduling can be accomplished by simply using a job to run such a script. -- Tim Young Elevate Software www.elevatesoft.com |
Tue, Oct 20 2009 9:26 AM | Permanent Link |
Leslie | Tim,
Glad I could help. Do you think you will come up with a little more sophisticated replication in the near future? Leslie |
Tue, Oct 20 2009 11:01 AM | Permanent Link |
Tim Young [Elevate Software] Elevate Software, Inc. timyoung@elevatesoft.com | Leslie,
<< Do you think you will come up with a little more sophisticated replication in the near future? >> Sophisticated in what way ? The only thing missing at this point is the ability to filter the updates for the target database. Other than that, you can do any type of replication possible. -- Tim Young Elevate Software www.elevatesoft.com |
Tue, Oct 20 2009 11:19 AM | Permanent Link |
Leslie | Tim,
I am tired, did not realize how my last post sounded. Sorry. It's my English. Usually I can express what I want to, but occasionally I screw it up. It was filtering what I meant. Leslie |
Tue, Oct 20 2009 5:58 PM | Permanent Link |
Tim Young [Elevate Software] Elevate Software, Inc. timyoung@elevatesoft.com | Leslie,
<< I am tired, did not realize how my last post sounded. Sorry. It's my English. Usually I can express what I want to, but occasionally I screw it up. It was filtering what I meant. >> No problem, I was just trying to clarify what you meant, and don't take offense at all to suggestions for improvements. -- Tim Young Elevate Software www.elevatesoft.com |
« Previous Page | Page 2 of 2 | |
Jump to Page: 1 2 |
This web page was last updated on Friday, May 3, 2024 at 08:07 PM | Privacy PolicySite Map © 2024 Elevate Software, Inc. All Rights Reserved Questions or comments ? E-mail us at info@elevatesoft.com |