Oliver,
Referring to the MySQL docs you were pointing me to: The way I
understood MySQL partitioning, there is *no* support for vertical
partitioning.
MySQL partitioning will allow you to distribute a single table over
multiple nodes. However, it will not route queries to different tables
on different nodes.
And: I was supposing that you would need distributed joins with
Sequoia even now. In case one table is in DB1 and another in DB2 and
you want both in a query. Isn't that so?
Yes, but this is not supported as it is mentioned in the limitation
section of the doc.
There was a prototype designed some time ago on top of C-JDBC, not
ported yet on Sequoia. But this comes at the expense of processing
queries at the middleware level which brings a significant overhead.
And: Don't you need an SQL parser in any case? How do you find out
which tables are involved in a query without one?
You need a parser but you don't need a full parser. Just extracting
table names can be done with a minimal parser made of regular
expressions. If you need to interpret the full SQL it becomes another
story, and the parsing overhead might become much larger.
Hope this clarifies things,
Emmanuel
--
Emmanuel Cecchet
Chief Scientific Officer, Continuent
Blog: http://emanux.blogspot.com/
Open source: http://www.continuent.org
Corporate: http://www.continuent.com
Skype: emmanuel_cecchet
Cell: +33 687 342 685
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia
