Can we use EntireX to replicate between Reptor on the mainframe and Adabas on Windows?
Why? We communicate to Adabas on the mainframe from other systems by using EntireX, for example .NET. Instead of dealing with two middle wares, we would like to deal with one only.
At this time, we are using WCP (Entire Network).
Thanks,
Dieter Storr
You can use EntireX but then you would have to write your own load program to pick up the records and store them in Adabas, because this is node to node replication and a target reptor can only reside on the mainframe.
So using Net-work is much easier because the reptor will create your Adabas calls for the target Adabas on Windows.
See also my questions under the forum EntireX.
SAG recommended to use EntireX over WCP to replicate Adabas from the mainframe to Adabas on Windows. SAG supports for other databases an adapter, for example SQL, but not for their own products. That’s unusual.
I experienced the following problems with WCP, for example
RC 148: Adabas nucleus not active – subcode 8224 - even ADABAS was up and running on the Windows side
RC 224: Entire Net-Work (WCP) – reply timeout
WCP0602E Node is unreachable
It may not active or there may be a configuration problem.
Detailed error response was: XTS1061E: TCP/IP Connection refused
If WCP stops working, I got a jam and an overflow in the replication engine (RPL buffer and/or SLOG).
It seems that the errors are caused by a bottleneck in WCP, when many files (100) were replicated - intitial state. At the end we will have more than 200 files replicated in production.
Organizations experienced better response and efficiency with EntireX, for example with RPC servers. I can agree with this.
WCP (Entire Net-Work) doesn’t give me the warm and fuzzy feeling to use in production.
I am not sure who recommended EntireX over WCP for this particular scenario.
On what arguments is this recommendation based?
I see good arguments using Net_work
Anyway using WCP is a tried and tested method for sending Adabas calls to different platforms including Windows.
It is certainly simpler in your scenario.
If something causes errors in this scenario you should contact SAG support and it needs to be fixed.
Of course there is some need to consider bottlenecks in such a scenario.
I would certainly recommend SLOGGING, particularly if you want to do Initial States to such a platform in parallel for many files.
You need a sort of buffer in case the target database is down or can not keep up with the laod.
Also newer versions of the event replicator have far better error handling in conjunction with Adabas to Adabas replication. (please see the documentation)
Also please note:
The product was designed based on market research that companies wanted to replicate from Adabas to relational databases.
In the beginning the original product assumed customers would do the load part for the target as the first big customer in fact did and we provid detailed information how we store data on EntireX.
Then we started to add Adabas to Adabas direct replication and produced a target adapter for relational databases.
We even provided node to node full support for Adabas to Adabas replication if done from mainframe to mainframe. It implies a source and a target reptor, which can exist (for good mreasons) only on the mainframe platform.
This was all based on market demand.
But the demand for replication to a target Adabas on non mainframe platforms is so far satisfied with WCP. Additionally you could use EntireX if and only if you were prepared to write your own load program.
I can not see where we do not support our own products.