Sizer enhancement

  • 27 January 2019
  • 0 replies

Userlevel 1
Badge +5
It has happened many times where I sized a cluster for customers according to their workloads (with N+1) and when I suggest HA the client looks at me with eyes wide open (after cluster deployment).

To my understanding sizing on N+1 only protects you from data loss. Should your original sizing be 75% or less on the dials (eg. 4 node config) you should be covered in the best effort scenario of a host failure. That's if the customer didn't add new workloads to exceed that "virtual" threshold. If all the dials were on eg. 90% you would still be safe on the data side but probably wouldn't see any workloads start that were running on the failed host.

Discussing HA with clients is a reality but not featured in sizer. I would like to see a tick box added in sizer in "sizing summary" that would effectively reserve enough compute and memory in the event of a host failure and reserved capacity to rebuild the data on the failed host.

If the client agreed to have HA and reserved storage capacity he can buy the cluster with easy of mind that all aspects and scenarios are covered. It will be easier for the person doing the sizing for his client to not have to keep thresholds in mind.

Setting the current thresholds in sizer is a manual process and mostly overlooked. Adding a tick box or radio button on the main page will make it a lot easier for the person doing the sizing exercise


0 replies

Be the first to reply!