Some general question with Clustering

[Christoph Gröver]

We are currently planning an application which may have to handle
a large amount of data.

So we started to experiment with MySQL and Clustering.

We have installed 5.0.24 on two storage nodes und on one SQL API node
with the management node additionally installed on the third computer.
(Each running SuSE 10.1, if it's important).

After reading many howtos and the official documentation I have still
a lot of questions left.

I'm trying to ask as specific as possible, so you may only have to ask with
a simple yes or no (if you'd include a link to a site with a possible no
it would be very nice).


So, what I've understood so far, is:

1. It's not possible to use NFS mounted filesystem to add large
amount of diskspace.


2. It seems each storage node has the complete data set (e.g. all
table information and the real content, but not the database
information).

3. The storage node cannot share data. So it is not possible to have
five 200 GB systems each having 1/5 of the data that is in total 1 TB in
size.
=> If we want to be able to handle 1 TB or more, each of the systems has
to have the ability to store this amount of data.


4. In MySQL 5.0.x the storage nodes save data to the filesystem, but SQL
statements are always run against the data that lies in the memory.
=> If we want to handle content of 4 GB, we have to supply the system
with 4 GB of memory (part of it can be swap space memory, which
makes statements run slower).

5. With MySQL 5.1.x it is possible (not with 5.0.x) to have tables
only in the filesystem. Queries are run over the filesystem, which is
slower, of course, but this way you can have hundreds of GB's of data.

=> So if we think that we will have to support a database size of a few
hundred GigaBytes of even 1 or 2 TB, we should definitively go for
MySQL 5.1.x ?

I tried to ask as precise as possible. Hope that you can easily
answer these. Thank you for your time.


Greetings from Münster