Typed Datasets - Performance over WebServices (1 Dst or Various Sm

[=?Utf-8?B?SXbDoW4gUMOpcmV6?=]

Hi there,
I have a question that has been raised on how an application should handle
the data that is travelling through a three-layered application.
Our data is going to be handled by Dataset object(s), and the data to be
handled is as follows:

We have N Companies with data, which is distributed among 1 table with all
the data that is similar on all Companies, and 1 table for each Company that
has specific data to store, so in case we want to query the db for some
company we would have to perform our query on those two tables, the main one,
and the specific of that company.

Suggestions has been made on:
1. Having 1 Dataset containing the main table and all of it's siblings
(knowing there will be N of them)
2. Having 1 Dataset containing the main table and 1 Dataset for EACH one of
the siblings,

Is it better to use a typed Dataset mirroring the SQL Server structure (All
tables and relations) or it might be better to send over the WebService 2
Datasets, one containing the non-specific fields and another containing the
specific data fields?
Looking for the best choice available, if there is another way to get this
data over the Webservice resulting in better performance it would be
appreciated.

Thanks in Advance, Iván

[=?Utf-8?B?UGV0ZXIgQnJvbWJlcmcgW0MjIE1WUF0=?=]

Depending on how much interoperability you want, you might want to consider
returning collections of objects (representing the datatables) instead. They
should be "slimmer" over the wire. The DataSet is a Microsoft - specific
animal that isn't readily understood by non-MS systems.
-- Peter
Recursion: see Recursion
site: http://www.eggheadcafe.com
unBlog: http://petesbloggerama.blogspot.com
BlogMetaFinder: http://www.blogmetafinder.com

"Iván Pérez" wrote:

Hi there,
I have a question that has been raised on how an application should handle
the data that is travelling through a three-layered application.
Our data is going to be handled by Dataset object(s), and the data to be
handled is as follows:

We have N Companies with data, which is distributed among 1 table with all
the data that is similar on all Companies, and 1 table for each Company that
has specific data to store, so in case we want to query the db for some
company we would have to perform our query on those two tables, the main one,
and the specific of that company.

Suggestions has been made on:
1. Having 1 Dataset containing the main table and all of it's siblings
(knowing there will be N of them)
2. Having 1 Dataset containing the main table and 1 Dataset for EACH one of
the siblings,

Is it better to use a typed Dataset mirroring the SQL Server structure (All
tables and relations) or it might be better to send over the WebService 2
Datasets, one containing the non-specific fields and another containing the
specific data fields?
Looking for the best choice available, if there is another way to get this
data over the Webservice resulting in better performance it would be
appreciated.

Thanks in Advance, Iván

[=?Utf-8?B?SXbDoW4gUMOpcmV6?=]

Hello Peter,
Thanks for your answer, i'm currently getting into the how to's of the
solution you suggested, main problem is, we currently have at our project (in
fact, on all the current projects at our company) a communication library
that works with stored procedures and returns the queried data into a
DataSet. Since using this library is mandatory, i'm fearing that i've to
stick to DataSets to return that data from the data-access layer to the
presentation layer.
If there is a way to manage that data-filled DataSet in the shape of object
collections, i would appreciate some guidelines on how to do it, if there's
no way on accomplishing this task, i'm still looking back at one of the old
solutions, but still don't know what is best to implement.

Regards. Iván

"Peter Bromberg [C# MVP]" wrote:

Depending on how much interoperability you want, you might want to consider
returning collections of objects (representing the datatables) instead. They
should be "slimmer" over the wire. The DataSet is a Microsoft - specific
animal that isn't readily understood by non-MS systems.
-- Peter
Recursion: see Recursion
site: http://www.eggheadcafe.com
unBlog: http://petesbloggerama.blogspot.com
BlogMetaFinder: http://www.blogmetafinder.com

"Iván Pérez" wrote:

Hi there,
I have a question that has been raised on how an application should handle
the data that is travelling through a three-layered application.
Our data is going to be handled by Dataset object(s), and the data to be
handled is as follows:

We have N Companies with data, which is distributed among 1 table with all
the data that is similar on all Companies, and 1 table for each Company that
has specific data to store, so in case we want to query the db for some
company we would have to perform our query on those two tables, the main one,
and the specific of that company.

Suggestions has been made on:
1. Having 1 Dataset containing the main table and all of it's siblings
(knowing there will be N of them)
2. Having 1 Dataset containing the main table and 1 Dataset for EACH one of
the siblings,

Is it better to use a typed Dataset mirroring the SQL Server structure (All
tables and relations) or it might be better to send over the WebService 2
Datasets, one containing the non-specific fields and another containing the
specific data fields?
Looking for the best choice available, if there is another way to get this
data over the Webservice resulting in better performance it would be
appreciated.

Thanks in Advance, Iván

[John Saunders [MVP]]

"Iván Pérez" <Iv*****@discussions.microsoft.comwrote in message
news:9E**********************************@microsof t.com...

Hello Peter,
Thanks for your answer, i'm currently getting into the how to's of the
solution you suggested, main problem is, we currently have at our project
(in
fact, on all the current projects at our company) a communication library
that works with stored procedures and returns the queried data into a
DataSet. Since using this library is mandatory, i'm fearing that i've to
stick to DataSets to return that data from the data-access layer to the
presentation layer.
If there is a way to manage that data-filled DataSet in the shape of
object
collections, i would appreciate some guidelines on how to do it, if
there's
no way on accomplishing this task, i'm still looking back at one of the
old
solutions, but still don't know what is best to implement.

Where in your layers are your web services? Are they between the
presentation and data access layers?

If so, and if you are satisfied to use a .NET-only solution, and if you are
using .NET 2.0, then go ahead and send the typed datasets between the
layers. On the other hand, do not add methods to your typed dataset on the
server using a partial class - the client will not be able to see the added
methods, properties, etc.
--
John Saunders [MVP]