Github user kl0u commented on the pull request:
https://github.com/apache/flink/pull/934#issuecomment-130700589
Hi @mxm ,
I totally agree that in terms of code clarity and structure, it would be
better to let all accumulators pass through the BlobCache. In fact, we also had
a brief discussion with @StephanEwen on the matter.
The only problem is the latency penalty that MAY result from this design
choice. In other words, if for most of the usages akka is enough, then forcing
small blobs to pass through the blobCache (i.e. reading and writing
accumulators to disk) may be expensive.
What do you think?
---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
