How to calculate search units in Azure AI Search ?

Narongrit Pooket 40 Reputation points
2024-12-02T21:49:36.4633333+00:00

How is the number of search units determined? According to this Microsoft documentation : https://learn.microsoft.com/en-us/training/modules/create-azure-cognitive-search-solution/2-manage-capacity , the number of search units is calculated by multiplying the number of replicas by the number of partitions (R x P = SU). However, Azure AI Search's S tier has 36 search units, despite having 12 replicas and 12 partitions. Why does the S tier have 36 search units instead of 144 search units ?

Azure AI Search
Azure AI Search
An Azure search service with built-in artificial intelligence capabilities that enrich information to help identify and explore relevant content at scale.
1,159 questions
{count} votes

Accepted answer
  1. Samantha St-Louis 115 Reputation points
    2025-01-10T00:43:10.52+00:00

    That's a great question!

    I'm going to give you two answers:

    1. The technical answer
    2. And an analogy because I think it makes these types of things easier to remember

    The technical answer

    R x P = SU (TRUE)

    BUT the S tier, you can configure up to 12 replicas and 12 partitions, but Azure caps the total active search units at 36, not 144.

    So why would tier S allows up to 12 replicas and partitions if it caps your SU's at 36?

    Because it allows combinations of replicas and partitions that total 36 SU or less while still allowing 12 partitions OR replicas as needed.

    For example:

    • 3 replicas x 12 partitions = 36 SU
    • 6 replicas x 6 partitions = 36 SU

    The goal is to max out performance and cost efficiency.

    The Analogy

    It's Hell's kitchen and there are 12 chefs (replicas) and 12 stations (partitions), but to make sure every chef can be properly evaluated only 36 active UNITS can be activated.

    • You can have multiple chefs sharing a station
    • Or one chef operating multiple stations

    Does that clear it up? 😊 If so please accept the answer so others can benefit from it as well!

    2 people found this answer helpful.

1 additional answer

Sort by: Most helpful
  1. Adharsh Santhanam 5,450 Reputation points
    2024-12-03T03:08:09.2133333+00:00

    Hello Narongrit Pooket, the discrepancy arises because the S tier in Azure AI Search uses larger partitions compared to other tiers. While you're correct in calling out the R*P=SU calculation, the S tier's partitions are significantly larger, which means fewer search units are needed to achieve the same capacity and performance. In the S tier, each partition is designed to handle more data and queries, so the effective number of SU is adjusted to reflect this increased capacity.

    Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.

    0 comments No comments

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.