Is it possible to have a redundant installation or something like that and implementing a high availability scenario for Harvest SCM Server? What if the Harvest Server and database crash? Is it possible to start a failover procedure starting up a redundant installation?
Some initial questions to consider:
- What operating system is the broker on?
- What type of database does Harvest use?
- Is the database on the same machine as the broker?
- What DR plan is already in place for the database? Or should the database be included in the same DR plan as the Harvest application?
A very simple fail-over scenario would involve installing the SCM Server software on a second backup machine, configuring the database connections so that it can connect to the same Harvest database as the primary, and making use of the "RT_FORCE_NODE_NAME" environment variable to "rename" the broker on this second machine to match the name on the primary. Then if the primary machine goes down you would
1. Set up an alias in the DNS directory pointing the SCM Broker name to the IP address of the backup machine
2. Start the broker on the backup machine.
This is not a complete plan, but a simple example. You should look at this as a starting point from which you could build to devise a plan that will work for your company and environment.
Information about the RT_FORCE_NODE_NAME environment variable can be found in chapter 15 of the Implementation Guide.
Let me know if this is what you needed.
Hi Melinda. Thanks for your Reply.
I give you the answers right now:
What operating system is the broker on? => AIX
What type of database does Harvest use? => ORACLE
Is the database on the same machine as the broker? => NO
What DR plan is already in place for the database? NONE - Or should the database be included in the same DR plan as the Harvest application? YES
Your suggestions looks good in case there is no problem with the database, and provides me certainty on what i was thinking about.
Anyway, it would be even better to know how to apply a similar scenario for a case in which the Oracle database crash too.
There should be another Oracle instance in another machine and another broker in another machine, when the crash occurs, it would be necessary to start up the whole secondary installation and connect that Harvest Server with the corresponding secondary Oracle Database, but how to be sure the secondary database has the same data as the primary? should the data be exported and then imported in some way? I need to know some pretty guidelines that helps me to write down a document that defines the procedure to take for implementing a High Availability scenario or even better, the procedure or document itself if it already exists.
Thanks in advance.
For Oracle database disaster recovery planning, the first person to check with is your DBA. He or she may already have a plan in place, and knowing that plan will help to tailor your plan for Harvest. It may be as simple as adjusting the parameters in the tnsnames.ora file to point to the backup database if the primary database goes down. If you do not have a DBA or they do not yet have a plan for recovery and fail-over, there is a lot of information on the Oracle website and the internet in general about example plans, templates and ideas, as well as consultants who might help you to set up a plan.
Your recovery plan should consider 3 possibilities:
1. the server that hosts the Harvest application goes down but Oracle database server is still up
2. The Oracle database server goes down but the Harvest application server is still up
3. Both the database server and the Harvest server go down at the same time.
Let me know if this makes sense.
Thank You Melinda.