(UCMDB) Support Tip – AM Adapter Pull vs Push
Old Population Jobs:
Asset Manager Pull:
•An old job is kind of draft. Uses internal customized generic adapter package, with nearly no support of AM mechanisms.
•It uses direct database access, not the AM library.
•It requires views to be created at database side, as only those view are accessed.
•Performance relatively good due to the technology used.
•Relatively painful to add support for new custom tables/fields.
•It creates locks on the database side when the job runs.
•This adapter is slow, with a lot of issues. It may take several releases to bring a correct functionality.
•It is based on the Asset Manager API, and can access to any table. It is as slow as the AM API are, requiring a fast database connection from the probe machine (ideally, the probe should be close to the am database in the datacenter).
•Multiple fixes have been implemented to solved big memory allocation issue, as result, it can now have a better performance.
•There is still many design bugs that needs to be addressed, such as poor logging, many timeouts without known causes, etc.
New Population Jobs:
Asset Manager Generic adapter:
•This adapter contains a rebranding of the push adapter, with small new internal features. It share the same comments/issues than the old push one.
•The population adapter is now fully using the AM API
•Population performance is lower than before due to the API, but still good enough for our needs.
•Push feature, is on the overall better than it was in the past.
•The standalone push adapter, may disappear in a few CP, as the generic adapter cover all needs for now, and they can be substituted easily.
•The generic adapter should be used for push and pull.
•Standalone push/old population adapters may be removed from the product, only the generic will remain.
•Population is by design faster than push, this won't change.
•Jump to latest releases is crucial for this adapter, as there is frequent fixes (refer to CP releases). It may be required to use last unpackaged AM libraries to solve some issues.
•Using population or push is mainly a question on where is the main data source.
•For AM license calculation, the push must be used.
•If ci are created on the am side, the generic adapter population is the way to go.
•If bidirectional sync is expected, use both.
If you find that this or any post resolves your issue, please be sure to mark it as an accepted solution.
Please hit the Kudo botton, if you find this post useful.