CODEWITHSUNDEEP
azureazure-data-explorer
Markus S.
Markus S.

Reputation: 2812

Azure Data Explorer Cost Estimator giving implausible estimates

I am trying to estimate Azure Data Explorer cost by using the suggested tool Azure Data Explorer Cost Estimator.

Entering the data below i get a total amount if $984 (if i add Kusto Markup fees which are only discounted until 17th May but we will on in Production beyond this date). Screenshot of Cost estimator with 0.01 TB collected per day

When i increase the value in the "Data collected (TB) per day" field to 0.02 the total price drops to $837.

Screenshot of Cost estimator with 0.02 TB collected per day

I understand, that the reduced price comes from suggesting only 2 Engine-VMs instead of 3 for the first amount of data. My assumption is, that 2 Engine VMs of Type D11 are more powerful than 3 VMs of E2A_v4 VMs.

In this case my plan would be to start with D11 Engine VMs and get more power for less money.

Is my assumption correct and does this plan make sense or is there anything i am missing?

Thanks for any suggestions.

Upvotes: 0

Views: 709

Answers (1)

Avnera
Avnera

Reputation: 7618

The VMs in Azure Data Explorer cluster provide two essentials capabilities:

  1. CPU (used to compute the queries)
  2. Disk (used to hold the hot data cache)

The E2a_v4 node has ~0.25 of the disk space and roughly equivalent CPU power compared to D11 node and it cost ~20% cheaper. So by trading 3 nodes of E2 with 2 nodes of D11, you are gaining more disk space (for the cache) but you lose 2 CPU cores used for computing the queries.

You should make the decision of which SKU to choose based on your workload. Assuming fixed budget, if the perf is good and the constraining factor is the disk space choose D11, if on the other hand you have small data size and can use more CPU, choose the E2a_v4.

Upvotes: 2

Related Questions