Re: Evaluating Xml vs. One field per row in vertical app
From: Michael Rys [MSFT] (mrys_at_online.microsoft.com)
Date: 01/28/05
- Next message: Michael Rys [MSFT]: "Re: How to query children of a given node?"
- Previous message: Irwin Dolobowsky [MS]: "Re: load both parent and child into one table!!"
- In reply to: Jerry Hayes: "Re: Evaluating Xml vs. One field per row in vertical app"
- Messages sorted by: [ date ] [ thread ]
Date: Fri, 28 Jan 2005 13:39:11 -0800
Did you take a look at some the SQL Server 2005 XML whitepapers on MSDN?
Note that in SQL Server 2005 the granularity is still at the row level and
not at the node level...
Best regards
Michael
"Jerry Hayes" <jhayes@NOSPAM.com> wrote in message
news:ey%23dDyBBFHA.2180@TK2MSFTNGP12.phx.gbl...
> Would security considerations be easier on the "field per row" approach?
>
> Seems that filtering out specific items might be more
> straightforward...not sure how this works on the schema level in 2005.
>
> Interesting topic, I'd like to see more discussion on the approaches.
>
> "Michael Rys [MSFT]" <mrys@online.microsoft.com> wrote in message
> news:uZBKEdABFHA.1400@TK2MSFTNGP11.phx.gbl...
>> Is this for SQL Server 2000 or can you play with 2005?
>>
>> In the later case, you can use the XML datatype for the flexible part and
>> promote some of the constant properties into the relational framework.
>>
>> HTH
>> Michael
>>
>> "M. Anders" <manders@nosmap.net> wrote in message
>> news:%23UlpzWABFHA.1404@TK2MSFTNGP11.phx.gbl...
>>> I'm designing a vertical app where the end uers field modification needs
>>> to
>>> be *extremely* extensive. Much more than a series of user-defined
>>> fields.
>>>
>>> Expected field count is around 25 fields, with key searches routinely
>>> done
>>> on about 5-8 of these.
>>>
>>> I'm evaluating an XML document design where the user can define their
>>> own
>>> schema vs. a design where I have a master record for the table with a
>>> limited static field set and keep the fields in a related table, one
>>> field
>>> per row.
>>>
>>> Ease of setup probably goes to the XML, as long as they don't try to
>>> modify
>>> the schema by hand and go way out of my expectations.
>>>
>>> A field per row is probably a little fatter initially, but as I audit
>>> changes, it gets leaner, where the XML would probably need to copy the
>>> entire document per edit.
>>> I'm thinking the field per row is probably a little better on searches
>>> too,
>>> taking advantage of all fields being indexed. Of course the queries
>>> aren't
>>> quite straightforward.
>>>
>>> Any thoughts or comments? Anyone done the same thing both ways that
>>> have
>>> stories to share?
>>>
>>> Thanks for your consideration.
>>>
>>> Mike.
>>>
>>>
>>
>>
>
>
- Next message: Michael Rys [MSFT]: "Re: How to query children of a given node?"
- Previous message: Irwin Dolobowsky [MS]: "Re: load both parent and child into one table!!"
- In reply to: Jerry Hayes: "Re: Evaluating Xml vs. One field per row in vertical app"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
