reduce the downtime in replicated environment [message #155746] |
Wed, 18 January 2006 05:42  |
guru_karnam
Messages: 142 Registered: May 2005
|
Senior Member |
|
|
Hello All,
I have a problem with replication.
I have two servers with ip addresses 131 and 132 respectively.
EACH SERVER HAS A DATABASE with schemas replicated in both servers.
If one server is down(primary),it switch overs to the second(secondary) and vice versa.
The tables and the data are replicated immediately, mean the downtime is very less.
But when i try to modify the metadata or make changes to the master group(group of tables,add or delete to the group), the changes should be reflected immediately in the second server.The downtime should be less when i make any changes in the master group of server1.During such scenarios, the tables of master group in the both servers gets locked and it takes more time to get replicated.
We decided to break the tables in the master group into smaller master groups so that the downtime is reduced.
Can anybody suggest me another way to solve this???
Regards,
RAVI
|
|
|
Re: reduce the downtime in replicated environment [message #155868 is a reply to message #155746] |
Thu, 19 January 2006 00:52  |
guru_karnam
Messages: 142 Registered: May 2005
|
Senior Member |
|
|
And one more thing, the data which is modified using DML commands is replicated immediately from one server to the other.
But when i make any alterations to the metadata of the tables present in the master group of one server, the changes are reflected in another server after 30 minutes or more (min 30 minutes).I want to reduce the downtime from one server to the other even if i make any alterations to the metadata.
Which is the best way to get a solution out of this?????
Regards,
Ravi
|
|
|