Re: Datasets with large amounts of data

From: mikeb (mike_at_nohostanywhere.com)
Date: 02/16/05 

Date: Wed, 16 Feb 2005 08:33:46 -0800

I have other ppc apps that use sqlce, I don't use datasets. I'm not big on
bound controls if this is what you're asking about. Might be that I'm an
old timer and like more control of what goes on with my data being
displayed.

Regardless.

The app is not a web services app.

The gist of the app is that they will download new shipment records from the
host server onto the ppc. they then go out to the shipping dock and scan
shipments into a container. When they scan the 'shipment' the app does a
lookup to find the given record and then flips a bit to denote that the item
is loaded onto a container. When complete, they take the scanner back to
the computer, and upload those records to the host.

Datasets are loaded when the users presses a Get Data button. We open a
connection to the host's sql server, run a stored procedure to get a
recordset and then load that data into the dataset table. When complete,
the sql server connection is disconnected.

Data is displayed on the ppc screen as retrieved from the dataset. Its a
scanner app, so the users scan ucc barcodes to start the lookup process.
When a barcode is scanned the app finds the appropriate record, flips a
Shipped switch in the dataset table record.

When all data is collected for the day (half-day, hour, whatever), they will
upload that data to the sql server for processing via a stored procedure. 

----
I'm trying to speed up the lookup process.  Right now, they can have up to 
5000 shipment records downloaded onto the handheld (into the dataset).  When 
they scan a barcode, the lookup can take a long time to find the right 
record.  If the shipment doesn't exist for the barcode scanned - this is the 
worst case scenario; the whole dataset is scanned.
So, my question is - would storing the data in a sqlce database help speed 
up this lookup process?  I would index the table on the UCC code that is 
scanned.
OR, can the dataset be indexed?  or is there some other method to do quicker 
lookups on a dataset?
---
does that help clarify?
Mike
"glenn" <ghan***@softeksoftware.com> wrote in message 
news:ePGjE0DFFHA.548@TK2MSFTNGP14.phx.gbl...
> I'm a little confused so sorry for the questions...
> 1) Is this a webservice type app where data is being queried and then
> transferred to the PDA to view or is the database actually residing on the
> PDA?
>
> 2) Are you using datasets to store the data in order to display in your 
> form
> elements?
>
> 3)  I'm a little confused about the comments of storing the data in
> sqlserver db and then asking if things would speed up if they were in
> sqlserver db.  Either way, the datasets are going to be the middle ground
> for storing the data for your components.
>
> What types of searching/sorting are you doing that is slow?  I think maybe 
> a
> little more info on what your app is doing might help provide a better
> answer for you...
>
> Sorry if I missed something that caused me to be more confused than
> necessary...
>
> glenn
> "mikeb" <mike@nohostanywhere.com> wrote in message
> news:uE$agmDFFHA.560@TK2MSFTNGP15.phx.gbl...
>> I came onboard to help finish a ppc application (vb.net cf) for a 
>> company.
>> The original developer is loading datasets from a sql server database for
>> data handling.   This worked pretty well (but very messy coding in my
>> opinion) up until we started dumping some real life data to the scanner
>> (3000-4000 records).
>>
>> With the large amount of data being loaded into the datasets lookups take
> an
>> exceptional amount of time to complete.
>>
>> My first question is - if we were to convert the datahandling method to
>> sqlce, would this resolve the speed issue?  Assuming we can index the 
>> data
>> properly.
>>
>> Second is, is there any way to speed up a dataset search?  They're using 
>> a
>> simple looking .select method to do the lookups:
>> ie. drow = dsVEKP.Tables(0).Select("UCC = 1234455566777")
>>
>> Can you index a dataset?
>>
>> Thanks!!
>>
>> Mike
>>
>>
>
>