Re: dataset Performence Issue



Adrian,

Call it personal opinion, but I am not a big fan of Prevayler*.* concepts. I
feel standard RDBMS's will evolve to provide that functionality.
But what I am certainly not a fan of is storing 400 MB in a DataSet - that
object is just not designed for such a heavy amount of in memory data.

- Sahil Malik [MVP]
ADO.NET 2.0 book -
http://codebetter.com/blogs/sahil.malik/archive/2005/05/13/63199.aspx
----------------------------------------------------------------------------

"Adrian Moore" <queryadataset@xxxxxxxxxxx> wrote in message
news:uSvzK%239lFHA.4000@xxxxxxxxxxxxxxxxxxxxxxx
> Sahil,
>
> There's nothing wrong with having a large, in-memory database. There are
> lots of reasons in a real-time environment where a database can fit in
> memory and performance rules. The SCADA system I work with has a 400 MB
> in-memory database. Data is flushed to disk, once an hour or on-demand.
>
> In-memory databases are also typically used in the embedded market since
> I/O to flash memory is usually not great.
>
> I see Datasets as a good solution to some problems.
>
> I see other open-source databases like Firebird, SharpSQL and SQLLite as
> good solutions when a small footprint RDBMS is needed. They are simple to
> install and setup.
>
> MSDE / SQL-Server Express is still overkill for many database needs, but
> does provide a great solution to problems that match it capabilities.
>
> I look foward to reading you book when its available.
> Ad.
>
>
> "Sahil Malik [MVP]" <contactmethrumyblog@xxxxxxxxxx> wrote in message
> news:ebYAid9lFHA.3552@xxxxxxxxxxxxxxxxxxxxxxx
>> If your DataSet will occupy 100's of MB, seriously that is ... ugghh ..
>> it makes me feel so pukish .. sorry but just don't do that. Such a big
>> dataset is like blasphemy, it's so awful, I am just so disgusted to even
>> hear that. Why do you need such a big dataset? That is an AWFUL way to
>> use a DataSet.
>>
>> In .NET 1.1, given the size of the dataset of course everything will
>> croak. Yes Especially GetChanges and Merge, both will be awful -
>> especially if you have lots of relations and a lot of tables.
>>
>> The workaround is - "A DATASET IS NOT A DATABASE, DON'T ABUSE IT AS ONE"
>>
>> .NET 2.0 has many enhancements under the scenes in both the various
>> underlying collections and the GetChanges algorithm and otherwise that
>> make using a DataSet a lot better, but 70MB? 100MB in a Dataset
>> (In-Memory Disconnected Cache?) - That is SUPER DUPER BUMPER ULTRA AWFUL.
>>
>> BTW, especially YOU need to read Chapters 9 and 10 of my book to
>> understand clearly and practically why I am advocating so strongly
>> against such a misuse/ABUSE of a DataSet.
>>
>> - Sahil Malik [MVP]
>> ADO.NET 2.0 book -
>> http://codebetter.com/blogs/sahil.malik/archive/2005/05/13/63199.aspx
>> ----------------------------------------------------------------------------
>
>


.



Relevant Pages

  • Re: dataset Performence Issue
    ... There's nothing wrong with having a large, in-memory database. ... to flash memory is usually not great. ... That is an AWFUL way to use ...
    (microsoft.public.dotnet.framework.adonet)
  • Re: Implement my own in-memory database
    ... I want to implement my own in-memory database. ... What resources would you suggest I use (online books, books to buy, ... I'd say a starting point would be a list of twenty or thirty conventional features / requirements, including stuff like programming interface, end-user interface, various logical capacities such as number of rows / tables per unit of available memory, language complexity, logical indexing requirements, memory management /storage efficiency,, whether multi-threading is really necessary, network support, blah blah blah. ...
    (comp.databases.theory)

Loading