Re: sql node and storage node concept, state of cluster.
Hi Qiang,
> everything i load to the sql node will store in storage node automatically?
The SQL node is a MySQL server (mysqld process). It provides the SQL interface to the storage nodes. Once you've created the database on the storage nodes and you've "nudged" all the storage nodes so they'll pick up all the tables (by running FLUSH TABLES), any data manipulation statements (SELECTs, INSERTs, UPDATEs, DELETEs) you execute on those tables via an SQL node participating in the cluster will be performed on the storage nodes.
> assuming i need to load the db into a new database world, i need to create the world database
> in sql node, then do the loading. what should i do after this ? go after EVERY storage node
> and create the new database world ?
Data definition statements aren't automatically distributed between storage nodes.
1. create the database and the tables on one storage node
2. create the database and run FLUSH TABLES on each of the other storage nodes
3. load the data via the SQL node
> what is the state of MySQL cluster? is it readly for production?
It is already being used in production.
> should i wait till the MySQL 5.1?
Since I don't know exactly what your requirements are, I suggest you see
http://dev.mysql.com/doc/mysql/en/mysql-cluster-in-5-0-and-5-1.html to get an idea of what's being added in 5.0 and planned for 5.1, then ask about specific features.
> is it stable now?
Only MySQL 4.1.7 and newer in the 4.1 series are considered production-ready. As always, we encourage you to use the latest production release in order to enjoy all available enhancements, security fixes and bugfixes, etc. Some people are using the 5.0.3 and 5.0.4 betas and finding them pretty reliable. (I'm currently running a Linux server with MySQL 5.0.3 and a Windows 2000 Server machine with MySQL 5.0.4, and they're both quite stable for me, but I'm not using them in a commercial/enterprise environment, either.) However, 5.0 is *not* yet guaranteed ready for production use. As for 5.1, you're welcome to download the source using BitKeeper, but there's no guarantee on any given day that it'll even compile.
> How much more advantage does it have comparing with replication with InnoDB engine?
Depends on what you need from it. Cluster currently does not support replication (we plan to add this in 5.1). On the other hand: you can already make backups running Cluster in MySQL 4.1; Cluster is transaction-safe; and unless you lose an overwhelming majority of the cluster's storage nodes all at once, your data will continue to be available.
Hope this helps answer some of your questions.
cheers,
jon.
Jon Stephens
MySQL Documentation Team @ Oracle
MySQL Dev Zone
MySQL Server Documentation
Oracle