Workload Management & Capacity Planning

Database
Enthusiast

Workload Management & Capacity Planning

I am seeking questions for followings question:

Q1.  If we have TASM activated on system with around 3 resource partitions and each having 5 Allocation group, is there any way we can calculate CPU seconds for a query for a particular AG in a particular Resource Partition? THE relative/known PS formula can gives us CPU share but how we can calculate CPU seconds from that?

Q2. If we have 3 active resource partitions (RP1,RP2 and RP3) and lets’ say on RP3  there is a CPU cap limit of (70%) will it, how we will calculate the CPU share for RP3 allocation groups and RP1 or RP2 allocation groups? Will it be the same formula to calculate CPU share or it will differ somewhat?

Q3. At what point of time we shall decide to increase the number of AWTs considering we have Reserved AWTs for Tactical Workload and we are facing circular queue situation continuously and quite often?

Q4. If we have to increase the AWTs then how we calculate this thing?

Q5. While doing capacity planning how we judge that how many nodes client need to buy more. Currently if its 60% data on all Amps and the perm space is expected to finish in 8 months’ time.

Q6. How we can calculate that client needs new nodes and how many?

Request: If some has capacity planning and forcasting material please share.

4 REPLIES
Enthusiast

Re: Workload Management & Capacity Planning


You have placed so many questions :). However, let me try to share few on database capacity planning based on your question so that in can  give evaluation or hint on cost-cutting.

It is just a cent :). There are so many things to consider

Hope you are in Teradata 14 where TVS keeps track of data temperatures. Temperature can be very hot, hot,warm and cold. There are various compression technniques like ALC, BLC,MVC and Autocompress.

You can suggest to client based on your studies and knowledge of the systems, whether you are going

for column values, logical compression for hash or join indexes, data blocks, column-partitioned compression.

In order to do this,you have to identify candidate database tables where you set the threshold current perm and if it is beyond this then compression is needed to save space. As you know,in ;Teradata, you can combine ALC,BLC, MVC in general but there are some exceptional cases.Besides you can have also row compression and autocompression

Few tips on MVC are:

Understand your data. Calculate the space used by incorporating compressed and uncompressed columns and then proceed with suggestions.

There are cases where data type and precisions;Have been used with higher than required. This has to be taken care of. Var[char,byte,graphic] if not frequently used can be used as LOBs.Number of compressible values does not affect Column width if upto 32 bytes.Calculate the BEP percent for 1,3,7,15,31,63,127,255 values on field size vs percent ;of rows that are compressible.

Cheers,

Raja

Enthusiast

Re: Workload Management & Capacity Planning

Thanks Raja for your response and time :)

But I am currently looking for answers in TD 12.

I would request if you can share some Capacity on demand and forecasting material with me on muhammadfahad.18@gmail.com.

That would be a great help thanks.

Re: Workload Management & Capacity Planning

Hi Raja,

If you have some material on Capacity Planning & forecasting Data, can you please share with me. You can mail me -

Sabyasachi00143@gmail.com

Thanks

Sabya

Re: Workload Management & Capacity Planning

Hi Raja,

I you have any material on the following, could you please send met at shahzadkasi@gmail.com

1. Performance Management and Query Optimization

2. Configuration/Space Management and Capacity Planning

Regards,

Shahzad Salam Kasi