[IPT] star schema functionality request

[Eli Agbayani]

Hi Tim, Markus,
Thanks for all your suggestions. We will try to see the best path for EOL.

Anyway, we've finished creating the EOL data object extension for IPT. We continued working on it because IPT can definitely still be used by many EOL providers who have good technical support and willing to undergo the IPT process.

We can send you the extension today.
Thanks.
Eli



----- Original Message -----
From: "Tim Robertson (GBIF)" <trobertson at gbif.org>
To: "Eli Agbayani" <eagbayani at eol.org>
Cc: ipt at lists.gbif.org, "Patrick Leary" <pleary at eol.org>
Sent: Sunday, May 10, 2009 5:16:03 AM (GMT-0500) America/New_York
Subject: Re: [IPT] star schema functionality request

Hi Eli,

A star schema for data transfer is indeed limited in scope, but offers a
lot more than is possible with a flat file, and supports many use cases we
face in the GBIF network related to serving occurrence oriented and
taxon/species oriented data.  Therefore I don't foresee the IPT handling a
more relational model than the star in the *near* future due to the
complexity of designing the user interfaces and output models to support
this.  Are you thinking of developing this functionality yourself using
the IPT as a base platform?  We can branch the source head if you get a
patch for this.

If you are looking to serve very normalised data, perhaps asking your
providers to support a standard exchange format is probably the best
option, which I believe EOL does now.  You might consider evaluating if
TapirLink supports a more normalised format, to provide a tool that
providers can install and map to your output model.

Best wishes,

Tim





> Markus, Tim,
> It is a concern for EOL usage of the IPT that the star schema is missing
> the functionality to add extension to another extension and not just
> always to the main DWC. The denormalized approach of doing actually this
> doesn't look good and providers will not like it.
>
> Another concern is the size of the recordset if we will (denormalize)
> repeat the main data object many times to accommodate many references,
> agents, audience type associated with it.
>
> We will continue testing IPT's usage for EOL while trying to minimize the
> overhead of the denormalized approach. But it would really be helpful to
> add the functionality making the star schema more relational.
>
> Thanks.
> Eli
>
>
>
> _______________________________________________
> IPT mailing list
> IPT at lists.gbif.org
> http://lists.gbif.org/mailman/listinfo/ipt
>
>


____________________________________________________________
Tim Robertson
Systems Architect
Global Biodiversity Information Facility Secretariat  (GBIF)
Universitetsparken 15, 2100 Copenhagen Ø, Denmark
http://www.gbif.org
trobertson at gbif.org
Phone: +45 35 32 14 87 (Office)
Fax: +45 35 32 14 80
____________________________________________________________