Hi all,
I am working on a project that uses Mysql.
My task is to provide a fault taulerant platform for it.
I would like some help to optimize the two ends of this platform.
There are few things that need to be solved.
High Availibility
Load sharing.
For HA, we are using a SAN disk to store the data and run heartbeat
between two nodes. This works fine. Only one node has access to the
disk and only one node can perform write operations.
The only problem is when/if heartbeat misbehaves or for some reason,
both nodes start to access the disk, there can be FileSystem
corruption(Not related to Mysql)
To avoid that situation, I would like to have a replica running that
can become master if neccessary or can be used to restore the data
to just before the crash state.
In future, I also want the first replica to provide replication
service to other application-server nodes so that they do read
operations on localhost and writes on the main server. (this part is
easy)
My design is something like this.
Machine1: primary for master-mysql
Machine2: Standby for master-mysql
master-mysql listens to 10.10.10.10 (floating IP for heartbeat)
Machine1: replicates from 10.10.10.10
Listens on 10.10.10.1
Machine2: replicates from 10.10.10.10
Listens on 10.10.10.2
These servers only grant Select access to the user
A Loadbalanced IP address 10.10.10.20 can be given to direct traffic
to 10.10.10.1 and 10.10.10.2 for SELECT queries. This way, we get a
little loadsharing.
I want other application server to replicate from the 10.10.10.20
(load balanced) IP address, but I think I will get into some
troubles with the binary logfiles naming. Will get to in in some
time when I readup more on replication.
So, now I would like some help on putting the correct settings on
the server that does the WRITES and the ones that do the READS.
I have at the moment for both(REPLICA and MASTER) of them the
following.
skip-name-resolve
skip-locking
sort_buffer=8M
read_buffer_size=2M
net_buffer_length=2M
key_buffer = 256M
max_allowed_packet = 1M
table_cache = 256
myisam_sort_buffer_size = 64M
thread_cache = 8
thread_stack=128K
query_cache_size= 16M
Both nodes currently have 1GB RAM but will be upgraded to 4GB in the
next few days.
Would you please advice, which settings I should remove from the
WRITE daemon and which to change for the READ daemon to get best
performance.
thanks in advance.
--
Atif Ghaffar
Tel: +41 (0)78 845 31 64
--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
