>>>>> "Nils" == Nils Swart <[EMAIL PROTECTED]> writes:
Nils> Hello,
Nils> I'm having a little problem with MySQL 3.22.13-beta.
Nils> The database i'm using is growing too fast, and is over 50 million records
Nils> at this moment (total size + indexes is approx. 5.5GB). As it has the
Nils> ability to grow to about 6.1 GB on this harddrive i want to solve this.
Nils> Does MySQL have the ability to spread the files over different disks ? eg.
Nils> the indexes and the data files spread over different paths (and different
Nils> disks)
Sorry, no. You can have indexes and data on different disks but not
spread a table over files.
(As all read/writes goes trough the functions 'my_write/my_read' it
would on the other hand not be that hard to solve this)
The major problem is that MySQL uses a 4 byte pointer to refer to a
row. With dynamic length rows, this efficiently restricts MySQL to 4G
rows. The new ISAM I am working on just now can use a 3-6 byte row
pointer and will solve this. If the OS supports it (for example
Solaris 2.6, Linux on Alpha..., BSDI) you can then access up to 48
bit files. Most systems also allows you to combine many hard disks
as one. I think this is the right way to go...
Nils> Furthermore, is it possible to create a database that has another 'home'
Nils> path then the standard $DATADIR MySQL uses ?
Yes. Just create a database anywhere and add a symbolic link to it in
the MySQL DATADIR.
Nils> Are these pure source-hacking jobs, or just configuration options I
Nils> didn't see ? Is it possible or planned that these features become
Nils> available in (near) future releases ?
The new ISAM will be available in MySQL 3.23.0
Nils> (Oh btw, i've thought of a RAID0 set, but the problem is that this DB will
Nils> be growing & growing.. I'd need to use some AMI RAID board which can
Nils> handle addition of disks after the RAID set has been made etc etc..)
MySQL has already an internal handler (MRG) that can access many
tables of identical layout as one. As soon as we get time we will
provide an interface to this and then you can simple store your data
in different files (for example per month) and access these as one
file.
Regards,
Monty
-----------------------------------------------------------
Send a mail to [EMAIL PROTECTED] with
unsubscribe mysql [EMAIL PROTECTED]
in the body of the message to unsubscribe from this list.
