Understand and optimize costs in Amazon Managed Service for Prometheus

The following frequently asked questions and their answers may be helpful in understanding and optimizing costs associated with Amazon Managed Service for Prometheus.

What contributes to my costs?

For most customers, metric ingestion contributes the majority of costs. Customers with high query usage will also see some cost based on query samples processed, with metrics storage being a small driver of overall costs. For more information about the prices for each of these, see Pricing in the Amazon Managed Service for Prometheus product page.

What is the best way to lower my costs? How do I lower ingestion costs?

Ingestion rates (not storage of the metrics) is the majority of costs for most customers. You can reduce ingestion rates by reducing the collection frequency (increasing the collection interval) or by reducing the number of active series ingested.

You can increase the collection (scraping) interval from your collection agent: Both the Prometheus server (running in Agent mode) and the AWS Distro for OpenTelemetry (ADOT) collector support the scrape_interval configuration. For example, increasing the collection interval from 30 seconds to 60 seconds will reduce your ingestion usage by half.

You can also filter the metrics sent to Amazon Managed Service for Prometheus by using the <relabel_config>. For more information about relabeling in the Prometheus agent configuration, see https://prometheus.io/docs/prometheus/latest/configuration/configuration/#relabel_config in the Prometheus documentation.

What is the best way to lower my query costs?

Query costs are based on the number of samples processed. You can reduce the frequency of queries to reduce your query costs.

To get more visibility into the queries that are contributing the most to your query costs, see Managing the query cost in Amazon Managed Service for Prometheus.

If I decrease the retention period of my metrics, will that help reduce my total bill?

You can reduce your retention period, however, this is unlikely to substantially reduce your costs.

For information about how to configure the retention period of a workspace, see Configure your workspace.

How can I keep my alert query costs low?

Alerting creates queries against your data, which add to your query costs. Here are some strategies that you can use to optimize your alert queries, and keep your costs lower.

In general, look at your alerting rules and make sure that you are alerting on the best metrics for your service. It's easy to create overlapping alerts on the same metrics or multiple alerts that give you the same information, especially as you add alerts over time. If you find that you often see groups of alerts happening at the same time, it's possible that you can optimize your alerts and not include all of them.

These suggestions can help you to reduce costs. Ultimately, you must balance the costs with creating the right set of alerts for understanding the state of your system.

For more information about alerting in Amazon Managed Service for Prometheus, see Managing and forwarding alerts in Amazon Managed Service for Prometheus with alert manager.

Can I check my bill at any time?

The AWS Cost and Usage Report tracks your AWS usage and provides estimated charges associated with your account within a billing period. For more information, see What are AWS Cost and Usage Reports? in the AWS Cost and Usage Reports User Guide

What metrics can I use to monitor my costs?

Metric samples you ingest are the primary cost driver for Amazon Managed Service for Prometheus. The number of samples ingested directly determines your monthly charges, making it essential to monitor and understand your ingestion patterns.

AWS Cost Explorer is the source of truth for monitoring your Amazon Managed Service for Prometheus costs. You can monitor Cost Explorer for historical and day-by-day trends for cost on Amazon Managed Service for Prometheus across multiple dimensions including ingested samples. AWS Cost Anomaly Detection can also provide you the ability to monitor unexpected changes in your spending patterns.

Using IngestionRate metrics provides an auxiliary method to monitor trends in ingestion that are directly correlated to cost. The advantages of using IngestionRate as an additional metric include:

For more information about monitoring Amazon Managed Service for Prometheus metrics in CloudWatch, see Use CloudWatch metrics to monitor Amazon Managed Service for Prometheus resources.

How do I view my costs in AWS Cost Explorer?

As your source of truth for Amazon Managed Service for Prometheus costs, AWS Cost Explorer provides your actual billed usage and charges for Amazon Managed Service for Prometheus samples ingested, including historical billing data by month and region. Use Cost Explorer for your final billed amounts and day-by-day cost trends.

To view your Amazon Managed Service for Prometheus costs:

How do I calculate the number of samples ingested in a month?

You can calculate the approximate number of samples ingested using Amazon CloudWatch's IngestionRate metrics with the AWS Command Line Interface. This is useful for reviewing monthly bills and understanding usage patterns across workspaces.

To retrieve ingestion data:

aws cloudwatch get-metric-data \
    --region your-region \
    --start-time start-timestamp \
    --end-time end-timestamp \
    --metric-data-queries '[
        {
            "Id": "e1",
            "Expression": "SUM(METRICS())",
            "Period": 3600
        },
        {
            "Id": "ws1",
            "MetricStat": {
                "Metric": {
                    "Namespace": "AWS/Usage",
                    "MetricName": "ResourceCount",
                    "Dimensions": [
                        {"Name": "Service",    "Value": "Prometheus"},
                        {"Name": "Resource",   "Value": "IngestionRate"},
                        {"Name": "Type",       "Value": "Resource"},
                        {"Name": "Class",      "Value": "None"},
                        {"Name": "ResourceId", "Value": "YOUR_AMP_WORKSPACE_ID"}
                    ]
                },
                "Period": 3600,
                "Stat": "Average"
            }
        }
    ]'

The command returns hourly average IngestionRate values, measured in samples per second. To calculate the approximate number of samples ingested in a month, multiply each hourly data point by 3600 (seconds per hour) to get the samples ingested in that hour, then sum all hourly totals across the month:

Monthly samples ≈ Σ (hourly IngestionRate average × 3600)

For example, if a single hour returns an average IngestionRate of 500 samples per second, that hour contributed approximately 500 × 3600 = 1,800,000 samples. Repeat this for every hour in the month and sum the results to get your approximate monthly ingestion count.

Key parameters:

To find your workspace IDs:

aws amp list-workspaces --region your-region

Use the workspace ID to filter the metrics to show data only for the specified workspace, rather than aggregating across all Prometheus resources in the region.

What data granularity is available for historical cost analysis?

Amazon CloudWatch's data retention policy affects the granularity available for historical queries:

For historical analysis beyond 63 days, CloudWatch automatically downsamples data to a minimum 1-hour period. When reviewing billing for months older than 63 days, you must use hourly aggregated data. The monthly sample calculation uses these hourly-averaged data points, summing each value multiplied by 3600 across the entire month.

This reduced granularity further contributes to why IngestionRate provides estimates rather than exact counts for older data—always refer to Cost Explorer for your authoritative billed amounts.

For more details on CloudWatch metric retention, see Metrics retention in the Amazon CloudWatch User Guide.

What are best practices for monitoring Amazon Managed Service for Prometheus costs?

To effectively manage and optimize your Amazon Managed Service for Prometheus spending, consider implementing the following monitoring practices:

Why is my bill higher at the beginning of the month than at the end of the month?

Amazon Managed Service for Prometheus has a tiered pricing model for ingestion, which results in costs in your initial usage being higher. As your usage reaches higher ingest tiers, with lower costs, your costs are lower. For more information about pricing, including ingest tiers, see Pricing in the Amazon Managed Service for Prometheus product page.

I deleted all my Amazon Managed Service for Prometheus workspaces, but I still seem to be getting charged. What might be happening?

One possibility in this case is that you still have AWS managed scrapers that are setup to send metrics to your deleted workspaces. Follow the instructions to Find and delete scrapers.