[
https://issues.apache.org/jira/browse/CASSANDRA-4175?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14222249#comment-14222249
]
Edward Capriolo commented on CASSANDRA-4175:
--------------------------------------------
There was once a https://twitter.com/roflscaletips suggestion that said
something to the effect of "make mongo faster by using small column names". The
same advice applies here. If you name a column "wombat_walnut_crackerjacks"
instead of "w" it is going to take up more space on disk. This is because
cassandra stores the column name and value each column on disk, because it is a
row store, apparently.
A simple way to solve this would be to have the CQL language store some
meta-data about alternate column names.
{quote}
Create table abc ( wombat_walnul_crackerjacks int (shortname w) );
{quote}
Then the query engine could allow either to be used in a select cause.
{quote}
SELECT w from abc;
{quote}
{quote}
SELECT wombat_walnul_crackerjacks from abc;
{quote}
An even easier way is to name the column "w". This way you avoid having systems
where column needs two names, or systems where column names have a internal
database of column name->shorter column name. But what is the fun of just
telling people to use short names when a complex solution can be engineered :)
> Reduce memory, disk space, and cpu usage with a column name/id map
> ------------------------------------------------------------------
>
> Key: CASSANDRA-4175
> URL: https://issues.apache.org/jira/browse/CASSANDRA-4175
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Jonathan Ellis
> Assignee: Jason Brown
> Labels: performance
> Fix For: 3.0
>
>
> We spend a lot of memory on column names, both transiently (during reads) and
> more permanently (in the row cache). Compression mitigates this on disk but
> not on the heap.
> The overhead is significant for typical small column values, e.g., ints.
> Even though we intern once we get to the memtable, this affects writes too
> via very high allocation rates in the young generation, hence more GC
> activity.
> Now that CQL3 provides us some guarantees that column names must be defined
> before they are inserted, we could create a map of (say) 32-bit int column
> id, to names, and use that internally right up until we return a resultset to
> the client.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
