Question
Nordea
SE
Last activity: 9 Sep 2020 10:14 EDT
Dual Database Setup
Hi,
Is there a possibility to have 2 databases in production (clone to each other) & application can auto-switch to 2nd database if the first is made down for some changes so that users do not experience any outage.
We are using Websphere for App server and Oracle 19c for Database
Accepted Solution
Updated: 9 Sep 2020 10:14 EDT
Proximus
BE
It's at DB server/Oracle level that this needs to be built.
It has to be built as a DB Cluster. 2 (or more) servers will be able to start an instance of your DB (full same DB) connecting to the same set of disks.
Then, your cluster mechanism will automatically start your DB on the Failover server if the master server is detected as down.
On our Prod system, this takes 1 minute in case the main server crashes.
You can have the same set of disks replicated on another data center with 2 others servers.
You've then got a cluster of 4 instances ready to be started as YourDB.
If your application supports 1 or 2 minutes without the DB, it can then be almost transparent.
Regards
Anthony
Pegasystems Inc.
US
I believe that is possible, but everything should be done transparent to Pega (at WAS/Oracle level), something similar to this: https://www.ibm.com/support/knowledgecenter/en/SSZJPZ_11.7.0/com.ibm.swg.im.iis.productization.iisinfsv.ha.install.doc/topics/wsisinst_haconnectclusterconfiguration.html
Nordea
SE
@KevinZheng_GCS I don't see any option of using an alternate server in the Websphere as mentioned in the URL above. However, it seems no such facility exists at WAS. Rather It may be achieved by using a load-balanced JDBC connection string from DB end where one DB can be failed over to another in case of any issue.
Accepted Solution
Updated: 9 Sep 2020 10:14 EDT
Proximus
BE
It's at DB server/Oracle level that this needs to be built.
It has to be built as a DB Cluster. 2 (or more) servers will be able to start an instance of your DB (full same DB) connecting to the same set of disks.
Then, your cluster mechanism will automatically start your DB on the Failover server if the master server is detected as down.
On our Prod system, this takes 1 minute in case the main server crashes.
You can have the same set of disks replicated on another data center with 2 others servers.
You've then got a cluster of 4 instances ready to be started as YourDB.
If your application supports 1 or 2 minutes without the DB, it can then be almost transparent.
Regards
Anthony