Should we consider SOA?

[patonp]

I've just come back from a discussion regarding a new project in which business users would like to execute DataStage jobs "on-demand". Although the jobs will likely be batch-oriented in their configuration (i.e. reading large volumes of data from static tables, transforming the data, then writing out to sequential files), the users may wish to run them at any time without having to contact the administrator.

One option that was raised was that of using the SOA edition of DataStage and executing the jobs via a web interface. Is this an appropriate use of the tool, or is the SOA edition intended for more of a message-based data stream? If the latter is the case, I'd appreciate any suggestions about alternative approaches!

Thanks in advance,

Peter

[ray.wurlod]

The problem with on-demand for a large batch of data is that the response time will be perceived as poor. I'd be more inclined to give users a command-style interface rather than to implement SOA for large volumes. Keep SOA for the short, snappy tasks, such as standardizing or certifying a row or two of data.

[patonp]

Thanks Ray - makes sense.

For future reference, if the volumes were lower (i.e. with a short run time), would SOA then be a good fit?

[ray.wurlod]

Yes. The rest of the answer is "how long is a piece of string?". Maybe if you implement the slow jobs first you can get some kudos by implementing the faster jobs later!

[patonp]

Thanks Ray!