TM1 Forum

I have a cube that has 7 dimensions and all its dimension are flat and carries only children elements. The cube carries quite some data and sizes to about 3.8GB (3,800,000 KB). I have 2 instances setup on the same server. One is the main, the other is fresh and has replication linked to the main.

When I create a replication of the cube, no data is copied over (Cube, views, rule, subsets did copy). The replication process does do its thing and takes over 5 minutes to complete, but once it is done, cube contains no data (.Cub file is 61KB). I seem to have no problem with any other cubes. I even created an identical cube with the same dimensions with less data and it worked fine. The only conclusion I came to was because my cube is too large. Is this a known limitation to the replication feature? Or perhaps a bug that needs to be addressed?

Any input is appreciated.

TM1 version 10.1.1 FP02

My personal opinion of replication is that should be avoided like the plague!! I wouldn't however just take my word for it. I would search on here for replication, I know there have been lots of post covering it,

Jim.

Kingsley wrote:I have a cube that has 7 dimensions and all its dimension are flat and carries only children elements. The cube carries quite some data and sizes to about 3.8GB (3,800,000 KB). I have 2 instances setup on the same server. One is the main, the other is fresh and has replication linked to the main.

When I create a replication of the cube, no data is copied over (Cube, views, rule, subsets did copy). The replication process does do its thing and takes over 5 minutes to complete, but once it is done, cube contains no data (.Cub file is 61KB). I seem to have no problem with any other cubes. I even created an identical cube with the same dimensions with less data and it worked fine. The only conclusion I came to was because my cube is too large. Is this a known limitation to the replication feature? Or perhaps a bug that needs to be addressed?

While I agree with Jim (not to the extent that Matt Goff agrees with Jim on this subject; he's both President and general secretary of the "We Hate Replication's Guts Society"), one thing that I would suggest you check is this:

Operation Guide wrote:•Transaction Logging - If you are performing a synchronization process, transaction logging must be enabled for the mirror cubes on the target server that are part of the replication and synchronization process. If you are performing a bi-directional synchronization, transaction logging must be enabled for all the related cubes on both the source and target servers.

TM1 uses these log files to keep track of the changes made to the source and mirror cubes. To verify that transaction logging is enabled, see Enabling and Disabling Transaction Logging in the IBM® Cognos® TM1® Operation Guide.

Check what your logging status is for the cubes.

Also there should be a replication log in your data directory. Check it to see whether there are any error messages recorded.

I have similar problem problem before. As our huge cubes have been turned off logging, replication doesn't work for us. I just create process to export and import. That is also not slow.

Cube logging must be switched on if you want the replication to be working. Cube logging switched off will simply end-up in not copying the data by replication. Replication relies completely on cube logs and timestamps of changes. I won't agree with avoiding replication, but I agree it has some pitfalls. For me the most annoying part is that automating replication can be done just through scheduling a chore. There is no way how to call replication from TI script as far as I know.