How to do GCP sizing for 100s of VM’s in 15 seconds?

Some times you just have to do sizing for Bill of Materials(BOM)/budgeting etc.

A lot of enterprise architects have to sizing and the public calculators given by the cloud providers just don’t make it any easier.

There is a one time 2 mins setup in any Linux/Windows/Chrome/Mac machine with Python2.x or Python3 and csv editing.

Setup:

  1. Use any linux/windows/mac machine to install required packages as directed in read-me.
  2. Download the files from here.
  3. And thats all!

Lets do pricing:

Modify the “pricing-input.xlsx”, sheet “Server List” and add the required values either with current vCPU & memory with utilisation or required sizing in target machines.

There is also a sample file in the input sheet for reference.

Samples and output from the video are also added to Git.

The output file has 7 entries for each VM it can do sizing for:

  1. hourly rate
  2. monthly no-SUD no-cud
  3. monthly with SUD
  4. monthly no-SUD 1yr CUD (Since there is no CUD for extended memory)
  5. Monthly_res_1yr monthly (with 1 year CUD)
  6. no-SUD 3yr CUD (Since there is no CUD for extended memory)
  7. Monthly_res_3yr(with 3 year CUD)

Choose the outputs from N1 or N2 runs.

Notes:

  1. If you are letting the tool to do sizing, it is targeting for 60% utilisation in the target state.
  2. At this time you can do sizing for N1 or N2 compute only by doing two separate runs. In the next update I will add E2, N2D etc.
  3. This configuration is not considering preemptible, small and Micro instances.
  4. All of the pricing is custom sizing only.
  5. This is indicative pricing tool that I built for myself and its not Google’s official sizing and pricing publication. I am using GCP published SKU’s and as far as I have tested its working within the 1% margin of error against GCP calculator.

GCP architect, AWS architect, CISSP, CCIE security. A learner. These days, learning to listen. Awful quality, but functional scripting skills.