>
>
> If the data is read from a slice of a partition that has been added over
> time there will be a part of that row in every almost sstable. That would
> mean all of them (multiple disk seeks depending on clustering order per
> sstable) would have to be read from in order to service the query.
If you make the timestamp the partition key you won't be able to do range
queries (unless you use an ordered partitioner).
Assuming you are logging from multiple devices you will want your partition key
to be the device id & the date, your clustering key to be the timestamp
(timeuuid are good t
If you disable compaction you will end up with a *lot* of sstables, this will
hurt read performance and be a pain to manage (including making repairs and
bootstrapping taking longer)
STCS is not too onerous, I’d recommend leaving on. If you want it to run less
frequently increase min_threshold.
The following article has some good information for what you describe:
http://www.datastax.com/dev/blog/optimizations-around-cold-sstables
Some related tickets which will provide background:
https://issues.apache.org/jira/browse/CASSANDRA-5228
https://issues.apache.org/jira/browse/CASSANDRA-5515
Whats your data model look like?
> I think it would be best to just disable compactions.
Why? are you never doing reads? There is also a cost to repairs/bootstrapping
when you have a ton of sstables. This might be a premature optimization.
If the data is read from a slice of a partition that
Hello Kevin
You can disable compaction by configuring the compaction options of your
table as follow:
compaction={'min_threshold': '0', 'class':
'SizeTieredCompactionStrategy', 'max_threshold': '0'}
Regards
Duy Hai DOAN
On Wed, May 7, 2014 at 2:55 AM, Kevin Burton wrote:
> I'm looking a
I'm looking at storing log data in Cassandra…
Every record is a unique timestamp for the key, and then the log line for
the value.
I think it would be best to just disable compactions.
- there will never be any deletes.
- all the data will be accessed in time range (probably partitioned
randoml
