> -----Original Message-----
> From: zhu hui [mailto:chinazhuhui04@stripped]
> Sent: Thursday, July 02, 2009 10:52 PM
> To: cluster@stripped
> Subject: helps needed for choosing a proper mysql cluster solution
>
> hi everybody
>
> I am a newcommer to mysql cluster area, nowadays deal with a problem
where
> we have mianly two big tables, and each have about 30 million rows of
data
> and performance
> was not so good, and as prediction, it will increase to 0.1 billion in
the
> coming two years.
This size should be within range of NDB. Just today I loaded 8M rows of
data into our cluster, using disk data storage, and it barely made a
dent in our capacity.
How big are the rows? NDB excels at handling simple data types. If any
BLOB columns are involved, it could be trickier, as these tend to stress
the throughput and capacity of NDB nodes.
> We hope to manage it in mysql cluster, or split it into two or more
mysql
> servers, but in the last method, we are afraid that there might be
some JOIN
> operations
> that needs to across serveral servers. so we hope mysql cluster can
cover
> all these for us.
What sort of operations do you perform on the tables? Transactions
involving single rows (or a few rows) are blinding fast on the cluster,
and scale very well. This load seems to be cluster's sweet spot.
Bulk operations and joins can be tricky. If you're looking for high
performance transaction processing, cluster may be for you. If you're
looking for more of a reporting or data warehouse application, you may
want to look at other options.
> I hope can partition each big table into serveral parts by its id, and
when
> data volume increases, we will add new server and it will handle it
> automatically, does that
> resonable in exist mysql cluster solutions?
The ability to add a data node online is available starting with NDB 7.
See
http://dev.mysql.com/doc/refman/5.1/en/mysql-cluster-online-add-node.htm
l for details.
Jeff
