Claudio Martella commented on GIRAPH-45:
Dmitriy, could you provide an example? We're looking for disk-basked
implementations that provide the usual append-only write performance and quick
scan-based read. For this reason i thought about sequence-file based approach.
I don't know any open source doing this without doing writes in-place
(tokyo-kyoto kabinet). I'm writing something like this
(memstore+sequencefile+compaction) pure java which is alpha-version now, I will
speed up dev in the future, but of course it won't be bullet-proof
[https://github.com/claudiomartella/sketches]. On this aspect there's a java
open-source implementation of leveldb, but don't know if it's stable enough for
us (we could take it (or sketches) under our umbrella thought)
Probably going default with something fully java (derby or something you're
going to suggest) and going pluggable is the best option.
I'll work on this from next week (moving to amsterdam next week), finishing
GIRAPH-10 right now.
> Improve the way to keep outgoing messages
> Key: GIRAPH-45
> URL: https://issues.apache.org/jira/browse/GIRAPH-45
> Project: Giraph
> Issue Type: Improvement
> Components: bsp
> Reporter: Hyunsik Choi
> Assignee: Hyunsik Choi
> As discussed in GIRAPH-12(http://goo.gl/CE32U), I think that there is a
> potential problem to cause out of memory when the rate of message generation
> is higher than the rate of message flush (or network bandwidth).
> To overcome this problem, we need more eager strategy for message flushing or
> some approach to spill messages into disk.
> The below link is Dmitriy's suggestion.
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
For more information on JIRA, see: http://www.atlassian.com/software/jira