If the EMS server is hosted on a multi CPU node, will using multiple file based datastores improve the EMS server performance?
Article ID: KB0086177
Updated On:
| Products | Versions |
|---|---|
| TIBCO Enterprise Message Service | - |
| Not Applicable | - |
Description
Resolution:
Yes, as the EMS server maintains a separate thread for each store file that is defined. By using N stores files, the EMS server will use N threads to read the datastore files upon start-up. Therefore, you should achieve the following performance benefits when using multiple stores by associating a destination with a store:
1. If you use different db files on different disk drives and you are using PERSISTENT messages, you should notice some performance increase though it might not be significant depending on the message volumes.
2. You could use dedicated stores files for heavy message volume s destinations. This may reduce store file fragmentation, size and EMS server memory usage.
3. During the recovery process, each defined store will read messages from a different thread. Some synchronization will occur but depending on how the stores are defined (different disks, etc.), you may achieve some performance improvement.
4. When the db files are compacted you may only need to compact one db file each time even though compacting one store does not allow destinations stored in other stores to continue their processing. However, the size of the store file may be much smaller than the size store file when you use default store files which will reduce the compacting time.
Environment:
= = = = = = =
ALL
Yes, as the EMS server maintains a separate thread for each store file that is defined. By using N stores files, the EMS server will use N threads to read the datastore files upon start-up. Therefore, you should achieve the following performance benefits when using multiple stores by associating a destination with a store:
1. If you use different db files on different disk drives and you are using PERSISTENT messages, you should notice some performance increase though it might not be significant depending on the message volumes.
2. You could use dedicated stores files for heavy message volume s destinations. This may reduce store file fragmentation, size and EMS server memory usage.
3. During the recovery process, each defined store will read messages from a different thread. Some synchronization will occur but depending on how the stores are defined (different disks, etc.), you may achieve some performance improvement.
4. When the db files are compacted you may only need to compact one db file each time even though compacting one store does not allow destinations stored in other stores to continue their processing. However, the size of the store file may be much smaller than the size store file when you use default store files which will reduce the compacting time.
Environment:
= = = = = = =
ALL
Issue/Introduction
If the EMS server is hosted on a multi CPU node, will using multiple file based datastores improve the EMS server performance?
Feedback
Yes
No
Powered by
