Re: dataset Performence Issue
- From: "Nigel Norris" <nospam@xxxxxxxxxx>
- Date: Thu, 4 Aug 2005 16:44:43 +0100
Just to get some facts instead of opinions, I ran some tests. The test
populated a DataTable with varying numbers of records up to 1 million.
The table was simple - 5 string columns, with one of those columns as the
primary key. Around about 100 bytes of data per row. The kind of application
I was thinking of was an address lookup table.
The two numbers I was interested in were:
- the load time per row
- the time to retrieve a row by primary key
Here's a few sample numbers (times in milliseconds):
Count Load Retrieve
1,000 0.02 0.009
10,000 0.03 0.013
100,000 0.04 0.018
1,000,000 0.06 0.028
CPU is an Athlon 2200. Framework version is V1.1
Basically the load cost per row does rise, but not too bad. However this is
a very simple table. I will interested to try this with the V2.0
improvements.
Retrieval time also rises, though quite slowly. Still, 30 microseconds per
lookup is probably better than you'll get out of a database.
So a reasonably large DataSet can work. However I would always be sure to
measure the behaviour with the largest amount of data that I wanted to be
able to handle before committing to it.
--------
Nigel Norris
.
- Follow-Ups:
- Re: dataset Performence Issue
- From: Sahil Malik [MVP]
- Re: dataset Performence Issue
- References:
- dataset Performence Issue
- From: Ashishthaps
- Re: dataset Performence Issue
- From: Sahil Malik [MVP]
- Re: dataset Performence Issue
- From: Adrian Moore
- Re: dataset Performence Issue
- From: Sahil Malik [MVP]
- Re: dataset Performence Issue
- From: Nigel Norris
- Re: dataset Performence Issue
- From: Sahil Malik [MVP]
- Re: dataset Performence Issue
- From: Nigel Norris
- dataset Performence Issue
- Prev by Date: Re: XML and dataset
- Next by Date: simple update method
- Previous by thread: Re: dataset Performence Issue
- Next by thread: Re: dataset Performence Issue
- Index(es):
Relevant Pages
|