cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Carl Yeksigian (JIRA)" <>
Subject [jira] [Commented] (CASSANDRA-7409) Allow multiple overlapping sstables in L1
Date Wed, 14 Jan 2015 18:22:35 GMT


Carl Yeksigian commented on CASSANDRA-7409:

I've pushed up an updated branch at:

The compaction selection has been updated so that it performs much better now. The biggest
issue was selecting a single sstable for the overlapping compactions, instead of the that
one plus the overlapping ones. It performs much better now.

|| || Operation Time || Compaction Time ||
| MOLO=0 | 3:52:46 | 0:21:04 |
| MOLO=2 | 3:45:52 | 0:37:50 | 
| MOLO=5 | 3:42:59 | 0:03:17 |
| LCS w/ STCS | 3:48:14 | 0:50:24 |
| LCS w/o STCS | 3:50:38 | 1:05:02 |

The performance on spinning disk is also improved by allowing overlapping; here are the results
of a read operation after running a large mixed read/write workload:

> Allow multiple overlapping sstables in L1
> -----------------------------------------
>                 Key: CASSANDRA-7409
>                 URL:
>             Project: Cassandra
>          Issue Type: Improvement
>            Reporter: Carl Yeksigian
>            Assignee: Carl Yeksigian
>              Labels: compaction
>             Fix For: 3.0
> Currently, when a normal L0 compaction takes place (not STCS), we take up to MAX_COMPACTING_L0
L0 sstables and all of the overlapping L1 sstables and compact them together. If we didn't
have to deal with the overlapping L1 tables, we could compact a higher number of L0 sstables
together into a set of non-overlapping L1 sstables.
> This could be done by delaying the invariant that L1 has no overlapping sstables. Going
from L1 to L2, we would be compacting fewer sstables together which overlap.
> When reading, we will not have the same one sstable per level (except L0) guarantee,
but this can be bounded (once we have too many sets of sstables, either compact them back
into the same level, or compact them up to the next level).
> This could be generalized to allow any level to be the maximum for this overlapping strategy.

This message was sent by Atlassian JIRA

View raw message