Based on throughput (number of connections and read and write requests) when resource is over provisioned or under provisioned, we recommend adjusting its type.

Recommendation is being triggere basing on the folloiwng conditions:

1. The account has more than minimal required throughput (400 RU/s per database).
2. The maximum consumption over the last week(hardcoded) is below 90%.
3. The account has at least one active database.

Different metrics are used:

• ProvisionedThroughput, actual throughput and max_consumption (Uses Azure Monitor API to get the metrics).

N/A.

Umbrella.

Identifying the resource as Idle if the number of requests is no more than 1 within the past 14 days.

Recommendation is being triggere basing on the folloiwng conditions:

1. The database is in a 'Succeeded' state.
2. The metrics (used list presented below) show minimal or no activity (max count ≤ 1).
3. The potential savings exceed the minimum potential savings > $5 as defualt.
4. The size properties can be successfully retrieved.

Different metrics are used based on database type:

• MongoDB: MongoRequests
• Gremlin: GremlinGraphDelete, GremlinGraphUpdate, GremlinGraphCreate, GremlinGraphThroughputUpdate
• Cassandra: CassandraRequests

1. Days To Check, default 14 days.

Umbrella.

Analyzing the usage on past 30 days and recommending to purchase reserved capacity if applicable, to optimize costs. The recommendation is being retrieved from the Azure advisor.

Recommendation is being triggere basing on the folloiwng conditions:

• 30-day period for usage analysis.

• Default recommendation: 1-year terms.

• Considers 3-year terms as alternatives.

Cost Calculation:

• Gets the current Cosmos DB costs.
• Filters costs by region and SKU display name (meter).

N/A.

Native + Umbrella validation.