TM1 Forum

Ok, here is what I have done.
1. Create a replication connection
2. Replicate some cubes and set to sync
3. Create a chore with replication connection
4. Run the chore
5. Successfully but from the log, only subsets and views have been replicated and no change on data

What have I missed?

Ok, after reading the posts inside the forum, I found that "Synchronize All Data" or chore looks at transaction log files in the source (Star) server to check if there are any changes, it will not be able to recognize those data changes that happened through TI Processes that are not recorded in transaction logs. Therefore you will loose the reconciliation of data between Star and Planet servers as you transfer data to Star through TI processes.
Since my cubes on Star doesn't have logging data through TI Processes, it won't apply any data changes on Planet cubes.
But from replicate interface of each cube, TM1 will immediatly replace the cubes and other objects after clicking on OK.
The working mechanism of this is kind of different from "Synchronize All Data".
That is my understanding and please correct me if anything is wrong.

macsir wrote:Since my cubes on Star doesn't have logging data through TI Processes, it won't apply any data changes on Planet cubes.

Normally, people turn logging off because they are loading from a source system, typically ODBC. If your planet servers can access that same source system then you don't need replication. If you've got a mix of budget and actuals and are trying to replicate the budgets to the planets then its a different story. What is driving your need to use replication?

rmackenzie wrote: Normally, people turn logging off because they are loading from a source system, typically ODBC. If your planet servers can access that same source system then you don't need replication. If you've got a mix of budget and actuals and are trying to replicate the budgets to the planets then its a different story. What is driving your need to use replication?

I just need to copy some cube's data for some proccess to run in another server. I can do export and import by text file but replication is quick and can sync all dimensions.

To use replication, you have to use the transaction log files.

rmackenzie wrote:To use replication, you have to use the transaction log files.

Ok, but why does not that replicate interface use the transaction log? It is just remove the original cube and replicate the new cube with data?

macsir wrote:Ok, but why does not that replicate interface use the transaction log? It is just remove the original cube and replicate the new cube with data?

Would you design a copy process (which is what replication is in the first place) that relied on transaction logging when the existence of said log is not guaranteed? Also, when you replicate a cube it has to replicate all the dimensions too doesn't it? How are you going to do that from a transaction log? The transaction log is needed for synchronization, which is the concept of keeping changes up to date on all affected servers. Replication and synchronization. although related by marriage, are two different things. If you think about it for a little bit it's all just common sense.

tomok wrote:

macsir wrote:Ok, but why does not that replicate interface use the transaction log? It is just remove the original cube and replicate the new cube with data?

Would you design a copy process (which is what replication is in the first place) that relied on transaction logging when the existence of said log is not guaranteed? Also, when you replicate a cube it has to replicate all the dimensions too doesn't it? How are you going to do that from a transaction log? The transaction log is needed for synchronization, which is the concept of keeping changes up to date on all affected servers. Replication and synchronization. although related by marriage, are two different things. If you think about it for a little bit it's all just common sense.

Understand, thanks, tomok. I was misled by thest two concepts. Replication and synchronization.