Scaling Factor in Workload modeling

I consider that scaling factor very important thing to consider during volumetrics analysis for designing workload model in performance testing.


The main resaon to consider Scaling factor is due to following reasons:

• Usually we know that NFT environment is always downsized to production
• The transactions that we cover in the scripts may differ from the transactions observed in the production
• Any other parameters like zone wise distribution of loads.

So considering all of the above facts simulating the exact load on production env needs a bit of calucalations to simulate the actual load.

Here is an example to understand:

We have following parameters in one of the application. in production we are having 1000 transactions our scripts covers 800 transactions and NFT env is half of the production.

So here ther are 2 scaling factors to consider .
1. NFT env which will be 0.5
2. transaction count which will be 1000/800=1.25

so the total SF is = 0.5*1.25=0.625

This is the number that we need to multiply in all hte individual trasactions to get the Load volume.

lets say we have 8 scenarios each will have 100 transactions  so each throughput 0.625*100=62

Here 496(62*8) transactions are the total throughput that we achieve in the load test.