Hello

1/ I have a search engine which crawls auction sites and returns
information which is inserted into a mysql database.  This activity
represents on average about 3 inserts per second with a combined payload
of 450 bytes per second added to the database.  This updating occurs
continuously throughout the day. Since this activity puts a burden on the
machine on which it occurs (not only the mysqld inserts but also the
scripts and http client activity), I am planning to dedicate one box to
the crawling/inserting and replicate the data to one (or possibly more)
other servers.  These slave servers will handle all client requests
(almost exclusively reads).

The mysql documentation states: "You should set up one server as the
master, and direct all writes to it, and configure as many slaves as you
have the money and rackspace for, distributing the reads among the master
and the slaves."

My questions are:

Since the updating occurs throughout the day and the same amount of data
has to be eventually inserted into the slaves I assume the updating will
require the same amount of resources (disk, cpu usage) on the slaves as on
the master - the same number of writes will occur on the slaves.  So the
the benefit of this configuration (in terms of performance) is that it is
the extra processing required to do the crawling will be offloaded to the
one master server, not the resources required for the mysql writes.  I
assume inserts done thru replication are not more efficient than regular
inserts. What is the performance hit of replication and is there a way to
limit the effect of the writes on the slave servers, thru configuration
parameters, for example, or would it even make sense to take the slave
offline at regular intervals while replication is taking place?

2/ Is there any documentation on handling and configuring large tables? 

Hope this is not too confusing... 
Many thanks

Todd Burke
phbnyc.com





-- 
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe:    http://lists.mysql.com/[EMAIL PROTECTED]

Reply via email to