pcd_au (8/2/2016)
Eirikur Eiriksson (8/2/2016)
You will have to understand how the SQL Server works, which workload is expensive and which factors bear more weight than others, data sizes and IO do have a very sizeable impact on performance.. Narrower data sets, narrower indices etc. do improve the performance to a greater proportion than the actual data size, especially when working on large analytical sets like DWs. I guess your lack of interest in the data size subject is portraying an ignorance on the overall performance subject.
😎
Interesting observation about my 'lack of interest in the data size subject', given you know nothing whatsoever about my background. I'll leave it at 'I started working with SQL Server in 1989, and have used it just about every working day since then.'. It's precisely my lack of ignorance about the subject that has promoted this question. Applications I develop tend to not have performance issues because I do understand the internal mechanisms of SQL Server, data modelling, indexing and performance tuning. It's the client that is making these requests and I'm looking for evidence that these things are important. Hence I'm reaching out to people such as yourself who have experience in developing sub optimal systems to get some metrics that I can take back to the client on why we should NOT restrict ourselves to a small subset of data types.
You input is much appreciated.
pcd
Not an observation but a direct response to your previous reply;-)
I'm not interested in '3-4 times reduction in size', but I am interested in 'large performance improvements by using the correct data types' - do you have any more info on this bit?
I have done this kind of optimization work many times and recently a reduction of a large data set to roughly 1/4 of the initial size, the performance improvement was 7-800 times improvement in the average response time.
😎