Hello fellows,
I would like to know if some one have experience about implement ADABAS CLUSTER SERVICE on a Production with several ADABAS nucleis which has its own physical database. Those ADABAS nucleis are on the same machine and one of them is used as a gateway; I mean there are structured in hierarchical way.
How can I implement a high-availability architecture using this kind of structure?
Best Regards,
I am not sure what your exact configuration is and what you want to find out.
But please note that Cluster Services is designed for a SYSPLEX environment with more than one machine (CPU).
It allows for several nuclei in different machines over the same database (DBID)
Running several nuclei in one machine (in a single LPAR) over the same database, Parallel Services would usually be a better match.
Rainer Herrmann
Hello Rainer,
Let me see if I’ve got a diagram of the customer architecture.
As you said ADABAS CLUSTER SERVICE is the solution that apply because the customer wants to implement an HA mechanism.
So the things is that they have several Adabas nucleis with differents DBID and one of them is used as a gate way, then using NSC goes to another DBID which has a FUSER to start the Natural application.
The customer architecture is very complex. As I said you before I will try to get the diagram of that Adabas structure.
Regards,
The structure you describe is fairly common: the FNAT and sometimes FUSER, FSEC are in different databases than the application data. Putting FUSER and application data in the same application database is a typical arrangement. Sharing the FNAT from one database is often used when there are multiple application databases - it allows the systems support staff to maintain Natural from one shared FNAT.
Parallel and Cluster Services use one DBID. So in this case, either the various databases need to be centralized in one database that is made more highly available or each of the databases will need to be made more highly available.
As Ranier says, unless the customer is using SYSPLEX, Parallel Services are probably the better route to higher availability.
Thanks a lot, it’s very helpfully