Re: [xml-dev] storing XML files

[Soumitra Sengupta <soumitra@b-bop.com>]

Hi Albena:

Please look at my answers right next to your questions.

Albena Georgieva wrote:

Stuff deleted.

> The general idea is storing several different data feeds, which arrive in
> XML format into a database. The data feeds can be very different (stock
> quotes, weather report, events information, daily news, horoscope, ...) and
> we receive them in XML format from different sources.
> Further, they are displayed on the WAP site, according to the search
> criteria of the user. Example: (select all the movies today in Amsterdam or
> show me the horoscope for THIS sign).
> After certain time the data feeds expire.
>
> Ideally, I would like to transform all of those different XMLs into some
> general
> format, from which sql output can be generated.
> I have though of SQL because the current database of use by the WAP site is
> SQL Server. The information is accessed through ASP files and store
> procedures.

This is a valid way of solving XML storage, indexing and retrieval problem if
you have:

a. you have pretty well defined DTD or schema for all the incoming documents and
they are unlikely to change.  You do not expect to add new DTD or schemas in the
future often.
b. the hierarchical relationship expressed in these documents are not very rich
or are not too nested.

If on the other hand you do not have control on the structure of the incoming
documents, they vary a lot and they are very rich, you should look at using a
native XML database.

>
> There is a slight possibility that the customer moves to Oracle, so my
> solution shouldn't be database provider dependent.
>

I recommend that you look at our Xfinity Server 2.0
(http://www.b-bop.com/products_xfinity_server.htm) as it will allow you the
benefits of a native XML database but protect your investment in RDMS technology
like MS-SQL server or Oracle.  It meets your requirement of "database provider"
independent.

>
> I have many, many questions, but the most frightening once are:

>
> Is it smart to develop a generalized format into which all those data feeds
> will be transformed (I hope an XSLT is a good tool to do that)? Or, I should
> deal with them individually?

Depends on whether the structure is pretty static, how many different structure
are you dealing with, are they likely to change a lot, are they very rich?  Also
how much time and money do you want to spend on development and do you want to
protect your existing investment?

>
> Can I apply XSLT on my XML data feeds and directly produce SQL statements?

I have not seen anything that does this.  Anyone with an answer to this?

>
> Has someone already developed an XML2SQL transformation tool?

> Just a flat
> table solution or I can also capture relationship information?

We have solved this problem in Xfinity.  All the rich relationships are captured
and stored.


>
> Is it the right thing to save those XMLs in to a relational database or
> shall I explore the native XML databases?

Depends on incoming XML and the needs of your application.

>
> Is there an existing DTD for SQL or that makes no sense to you?
> My system should be flexible enough to accept new kinds of data feeds ...
>

This I think is the key requirement that you have.  Unless there are other
requirements that I do not understand or know of, you really should look at
native XML databases.

>
> Hope, I made my worries, doubts and knowledge holes clear.
> Any suggestion is welcome.
>

Do not hestiate to ask questions.  This group is a great place to learn and you
will find that people are actually quite nice and helpful.

Regards,

Soumitra