I’m wondering whether a query for a huge amount of data on Tamino under SuSE Linux (mf)could be done against Adabas C under OS/390. In this case, both products would be installed on the same server.
Have your ever tried this? Is this better than Tamino under Windows?
the evaluation of a Tamino query against an X-Node Adabas can roughly
be summarized as follows:
- assume the query is something like “Employee[Name = ‘John’]”
where “Employee” is mapped to an Adabas file, and “Name” is mapped
to an appropriate field in that file,
- the filter [Name = ‘John’] is transformed to the corresponding
Adabas search buffer which is then evaluated by Adabas,
- the result is then sent to Tamino which wraps the Adabas records
into XML documents.
This means the expensive parts are the network connection for the
data transport (which depends on the result size), and the CPU
consumption for the wrapping. If the network is fast enough I’d
run Tamino on the faster machine.
If both systems run on the same machine, of course some general
aspects also apply:
- there should be enough main memory so that both systems can
run with convenient buffer pool sizes,
- the databases should reside on separate disks.