Re: Huge data
From: Jon Skeet [C# MVP] (skeet_at_pobox.com)
Date: 10/06/04
- Next message: localhost: "Index DataColumns In Advance?"
- Previous message: Patrice: "Re: changing connection properties after opening when pooling is enabled"
- In reply to: jy: "Re: Huge data"
- Next in thread: Sahil Malik: "Re: Huge data"
- Reply: Sahil Malik: "Re: Huge data"
- Messages sorted by: [ date ] [ thread ]
Date: Wed, 6 Oct 2004 15:38:22 +0100
jy <jy@discussions.microsoft.com> wrote:
> > Well, you *could* use DataReader, but I wouldn't recommend that. Just
> > use a select statement that selects a single row, or a page-worth of
> > rows.
>
> I haven't much used dataReaders. Can you tell me how to select a group of
> rows and sequentially call the next group when needed?
Doing that would be a bad idea, IMO. You really don't want to use
DataReader here.
> > > 2- won't it be too slow?
> >
> > Have you tried it? We don't know much about your database - you haven't
> > told us much information about the situation.
>
> Well, the database is made of over 1000 Tables. You have around 50 users.
> It's an Oracle 9i DataBase. I haven't tried it but I suspect that having to
> connect everytime to the database to get a record or a group of records will
> take more time than retreiving all the data and work with it offline, no?
Not necessarily - and if working with it offline just isn't an option,
as it seems it isn't, given the exception you've got, surely it's
better to have something that works than something that doesn't.
Note that connecting every time doesn't necessarily mean making a new
underlying database connection each time - connection pooling takes
care of that for you.
> > > 3- How will I search the tables for specific rows?
> >
> > Using a select statement.
>
> Yes, but by using select statement, I will lose the feature of
> DataViewManager to filter locally the rows and displaying the rows I need at
> any time. Plus, I will have to send a query to the database everytime.
And do you have evidence (rather than just guesses) that that will be
too slow?
Don't forget that you could always fetch, say, 100 records and then
allow further filtering within those.
> > > 4- Will it be possible to update / delete / insert rows?
> >
> > Yes. For update/delete, you use a row you've already fetched. For
> > insert, you don't need any existing data, do you?
>
> Yes but it won't be working like the rest of the application where I could
> update many rows before send the commit to the database.
So keep a table with all the data you're updating and then send it in
one go - that still doesn't mean you need to have the whole table in
memory at a time, unless you're updating every row in the table.
> In a way, I know that all these arguements are minor problems, but I need to
> know if it's possible in this case to make it work like the rest of the
> application (DataViewManager filter, multiple update before commit, ...), or
> if it will have to be different.
If you can't fit the whole of the data into memory, as it looks like
you can't, then it will clearly have to be different.
-- Jon Skeet - <skeet@pobox.com> http://www.pobox.com/~skeet If replying to the group, please do not mail me too
- Next message: localhost: "Index DataColumns In Advance?"
- Previous message: Patrice: "Re: changing connection properties after opening when pooling is enabled"
- In reply to: jy: "Re: Huge data"
- Next in thread: Sahil Malik: "Re: Huge data"
- Reply: Sahil Malik: "Re: Huge data"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
