Assigning two DS engines to the same xmeta repository

jinm · Post by **jinm** » Wed Dec 12, 2012 6:57 am

Hi guys
Want to know if anyone has experience in using the same metadata repository for two (or more) DS Engines.

That is when you install the second engine, then you should be able to select an already existing db instance to work as repository for jobs running the second engine also, eliminating the option to run create_tables.sql etc.

If this is not possible, then a brief explanation of why not??

Searched all over, but apparently there is no ocument anywhere describing this setup (did however find a drawing at the publib site from IBM

ray.wurlod · Post by **ray.wurlod** » Wed Dec 12, 2012 12:48 pm

That's the way it's intended to work. The last two sites I've installed have five and two engines respectively. You install selecting Engine tier only, and are then asked for the location (and other information) for the Services tier.

chulett · Post by **chulett** » Wed Dec 12, 2012 1:46 pm

Meaning you can share the repository? Or cannot? Sorry but I can't tell what the intention is you are trying to communicate.

stuartjvnorton · Post by **stuartjvnorton** » Wed Dec 12, 2012 4:37 pm

Isn't Ray just saying that you can have a number of Engine tier machines connecting to the same Services tier (and therefore repository)?

chulett · Post by **chulett** » Wed Dec 12, 2012 4:57 pm

You tell me.

ray.wurlod · Post by **ray.wurlod** » Wed Dec 12, 2012 6:16 pm

Planning, Installation and Configuration Guide version 8.7 page 22 Figure 14 - Topology with Two Engine Tiers

For "two" read any reasonable number of Engine tiers.

jwiles · Post by **jwiles** » Wed Dec 12, 2012 8:54 pm

The engines can register to the same services tier, and would store their respective metadata within the same repository database. Only shared metadata, as defined by Information Server (shared table definitions, etc.) can be accessed by both engines, but they would NOT share jobs and other engine-specific objects.

Regards,

chulett · Post by **chulett** » Wed Dec 12, 2012 10:41 pm

Thanks... that was the clarification I was looking for.

jinm · Post by **jinm** » Fri Dec 14, 2012 1:15 am

Thanks a ot for the effort guys.

"but they would NOT share jobs and other engine-specific object" - which is my problem.

I think we will go with a solution to have two repositories, located in the same physicla database under different schema owners.
Oracle will than be able to handle high availability on the repository

BR Jan

ray.wurlod · Post by **ray.wurlod** » Fri Dec 14, 2012 3:32 am

That "solution" will still not share jobs and other engine-specific objects.

Information Server does have built-in HA capability in the services and repository tiers, but there's no way to get running jobs or connected client sessions to fail over.

zulfi123786 · Post by **zulfi123786** » Fri Dec 14, 2012 3:56 am

Two engines on same database and different schemas ? or two different databases in same instance ?

The last time i was trying to install 8.5, it was asking for a fresh database.

lstsaur · Post by **lstsaur** » Fri Dec 14, 2012 12:15 pm

No, for Oracle there is no such thing as "two different databases in same instance." Are you talking about MS SQL Server?

jinm · Post by **jinm** » Mon Dec 17, 2012 1:42 am

Issue is all about saving money - Running one Oracle database costs x thousand euros a year.
My idea was that if fail over engine and services could use the same Oracle repository and leave the Oracle part of the failover to say Data Guard, than a pretty amount of money could be saved.

Instead we now use different schemas in the same Oracle Instance for the two sets of engine/services, and then save the license fees in that way.

Automatic failover of running jobs etc is not something we request or desire