[mdr-users] Possible usecase?

[Mick Wever]

I've got a usecase that i'm wondering would suit being implemented with
MDR.

I've searched the mailing lists, and read the doco.
To my defense it's a common statement that there isn't enough doco,
examples, usecases to really understand what MDR can be used for and to
what benefits it will give...

The usecase is for a search engine.
I would like to record every query that is submitted to the search engine.
Most importantly I would like to start associating the metadata relating
to each query. Timestamps for each submit, analysis scoring against the
different backend search indexes, which results were viewed by the user,
etc, etc.
This data would come from different sources, eg apache access_logs, log4j
logs, or even directly from the search engine.

The obvious end-use for such a metadata repositiory would be statistical
reporting. But it would also be nice to use for inputting *real* data into
a testing framework (eg testng). It could even have applications directly
by the search engine.

It must be a central repository, therefore I imagine using the jdbc
storage, that is accessible from many different hosts. While write
operations do not need to be performance critical, read operations will be.


Have I understood MDR correctly and is this a realistic candidate for MDR?
Any there other frameworks more applicable?

Are the suggestions or more in-depth examples I can read relating to
any of the requirements mentioned?
Does anyone have hints on avoiding possible problems implementing such a
usecase?

Will any of the mentioned requirements need to be submitted enhancements
to MDR?

What is the performance of JDBC storage like?
Are there size limitations of MDR, will it handle *gigabytes* of data?

mck

[Petr Hrebejk]

Hi,

From my point of view, what you are doing is a plain database. Not everything
what people call metadata is "meta" in terms of MOF. But I of course may be wrong.

What specific advantages you see in using MDR for your purposes instead of
normal database? Are you familiar with MOF and JMI? What metamodel are you
thinking about to use? Own? Existing?.

Cau
H

Post by Mick Wever
I've got a usecase that i'm wondering would suit being implemented with
MDR.
I've searched the mailing lists, and read the doco.
To my defense it's a common statement that there isn't enough doco,
examples, usecases to really understand what MDR can be used for and to
what benefits it will give...
The usecase is for a search engine.
I would like to record every query that is submitted to the search engine.
Most importantly I would like to start associating the metadata relating
to each query. Timestamps for each submit, analysis scoring against the
different backend search indexes, which results were viewed by the user,
etc, etc.
This data would come from different sources, eg apache access_logs, log4j
logs, or even directly from the search engine.
The obvious end-use for such a metadata repositiory would be statistical
reporting. But it would also be nice to use for inputting *real* data into
a testing framework (eg testng). It could even have applications directly
by the search engine.
It must be a central repository, therefore I imagine using the jdbc
storage, that is accessible from many different hosts. While write
operations do not need to be performance critical, read operations will be.
Have I understood MDR correctly and is this a realistic candidate for MDR?
Any there other frameworks more applicable?
Are the suggestions or more in-depth examples I can read relating to
any of the requirements mentioned?
Does anyone have hints on avoiding possible problems implementing such a
usecase?
Will any of the mentioned requirements need to be submitted enhancements
to MDR?
What is the performance of JDBC storage like?
Are there size limitations of MDR, will it handle *gigabytes* of data?
mck