0

I need to consolidate data from 30 databases and 20 tables. My first attempt was to create one data flow for each table. Each data flow would have 30 db sources which takes data from 1 table from each db and then inserts in 1 destination. This doesn't work because there is no enough memory. I tried to filter data to ignore data that's not going to be used - it didn't help. These tasks should perform on daily basis. Whole Consolidation db is deleted and then new data is inserted, which means that huge amounts of data are inserted daily. Anyone has a suggestion of how to handle memory problem?

The only solution for now I see is to create 20 packages. Each package would have 30 data flows. Each data flow would take data from only one database. But then again, I don't know what is the order of execution? Will data flows of one package start to execute at the same time or they go one by one?

EDIT:

This is how I did it... here's screenshot of Control flow: enter image description here

Screenshot of Data Flow: http://img833.imageshack.us/img833/355/45908206.png

I created one package for one table from all databases. In this case, each dataflow takes 4-year old data from a table and imports to a source database. The problem is that I have one table in one database which has too much data and it returns error:

The buffer manager failed a memory allocation call for 10484608 bytes, but was unable to swap out any buffers to relieve memory pressure. 20 buffers were considered and 20 were locked. Either not enough memory is available to the pipeline because not enough are installed, other processes were using it, or too many buffers are locked.

Any suggestions?

Brian Tompsett - 汤莱恩
  • 5,753
  • 72
  • 57
  • 129
ilija veselica
  • 9,414
  • 39
  • 93
  • 147

1 Answers1

2

Here is what I would suggest if you are planning to do this in SSIS.

  1. Let's assume that you have tables A, B, C ..., R, S, T in databases DB1, DB2 ... DB30. And also your consolidation database in MainDB.

  2. Create an SSIS package that will contain 1 or more data flow tasks to transfer data from tables in database DB1 to tables in database MainDB. To do this, you can use a OLEDB Source to retrieve data from DB1. If you have to apply some transformations, you can do that with appropriate transformation tasks. To insert data into destination tables, you can use an OLE DB destination.

  3. Save the connection string in package configuration file (dtsConfig)

  4. Create 30 different configuration files each containing the connection string for the 30 databases that you have.

  5. You can create 30 SQL jobs under SQL Server Agent to schedule the daily run.

  6. Each of the SQL jobs will run the same SSIS package but make use of the corresponding configuration file to point to the correct database.

Other options are:

  1. Using SQL Server Replication. You need a two step process. Truncate the data in MainDB using an SQL job and then sync the data between the databases.

  2. Use commercial products like Redgate or Idera

Hope that gives you an idea.

  • "Create an SSIS package that will sync the 20 tables between database DB1 and MainDB." - What do you mean exactly by "sync"? I don't understand this step. Thanks for the answer! – ilija veselica Aug 23 '11 at 10:17
  • I have edited my question... the problem is in memory, obviously this method tries to handle too much data. Thanks for your help! – ilija veselica Aug 29 '11 at 10:25
  • there are few details missing in this answer to make it understandable for a beginner... here is what I've came to in the end: 1) This is the method I've used in data flow task to insert/update/delete records: http://stackoverflow.com/questions/7254079/ssis-delete-rows 2) Using FOREACH loop: http://stackoverflow.com/questions/7347284/ssis-using-foreach-to-read-from-multipe-dbs 3) If you run into memory problems, this could be on of the solutions: http://stackoverflow.com/questions/7393027/ssis-out-of-memory-error-again – ilija veselica Sep 15 '11 at 12:32