Github user mengxr commented on the pull request:
https://github.com/apache/spark/pull/1687#issuecomment-50785472
I meant the final `userFeatures` and `productFeatures` stored in the matrix
factorization model. If those two RDDs are kicked out from memory by later
jobs, we have to restart from the very beginning. Having more partitions can
also help lookup. So I'm thinking about changing
https://github.com/srowen/spark/blob/SPARK-2768/mllib/src/main/scala/org/apache/spark/mllib/recommendation/ALS.scala#L287
to
~~~
usersOut.setName("usersOut").persist(StorageLevel.MEMORY_AND_DISK)
productsOut.setName("productsOut").persist(StorageLevel.MEMORY_AND_DISK)
~~~
and maybe also make them have more partitions in `unblockFactors` for quick
lookup.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
