Stay organized with collections Save and categorize content based on your preferences.
Jump to

Managed Service for Prometheus

A fully managed and easy-to-use monitoring service, built on the same globally scalable data store used by Google Cloud.


Use Prometheus without managing infrastructure

Fully managed Prometheus®-compatible monitoring stack with default two year retention and global queries over regionalized data. No need to federate, add resources manually, or devote time to maintenance.

Default two year metrics retention included

Sharding an expanding storage footprint is a hassle when running your own Prometheus-compatible aggregator. To alleviate this pain point for you, all metrics are stored for two years at no additional charge.

Keep open source tooling and avoid vendor lock-in

Maintain compatibility with PromQL so you can keep your preferred open source tools like Grafana®. Configure deployment and scraping via any open source method, such as prometheus-operator or annotations.

Key features

Key features

Backed by Monarch, Google’s in-memory time series database

Managed Service for Prometheus uses the same technology that Google uses to monitor its own services, meaning that even the largest Prometheus deployments can be monitored at global scale. Since this is the same back end that powers Google Cloud’s monitoring service, you can query, visualize, and analyze metrics from both services together in Cloud Monitoring.

Use Cloud Monitoring with Managed Service for Prometheus

You can view Prometheus metrics and Google Cloud system metrics together for a “single pane of glass” across your infrastructure and applications. Managed Service for Prometheus is built on the same technology and backend as Cloud Monitoring, so your Prometheus metrics can be used with the dashboarding, alerting, and SLO monitoring available within that service. Chart your Prometheus metrics right alongside your GKE metrics, your load balancer metrics, and more.

Managed or self-deployed collectors

Managed Service for Prometheus offers managed collectors that are automatically deployed, scaled, sharded, configured, and maintained. Scraping and rules are configured via lightweight custom resources (CRs). Migration from a Prometheus operator is easy, and managed collection supports most use cases. You can also keep your existing collector deployment method and configurations if the managed collectors do not currently support your use case. Onboarding is as easy as copying your existing deployment configs and changing the container image, and you can keep running your existing Prometheus stack alongside Managed Service for Prometheus.

View all features


Customers are freeing up developer time and keeping their open source tools

Horizon Blockchain Games logo

"We have been running Prometheus ourselves for GKE metrics, but the ongoing maintenance took up too many development hours. We started using Managed Service for Prometheus and it just works. It can handle whatever volume we have because it's built on the same back end that Google uses itself, and we get to keep using the same Grafana dashboards as before while keeping open standards and protocols."

Peter Kieltyka, CEO and Chief Architect, Horizon Blockchain Games



Documentation overview

Get started with Managed Service for Prometheus.

Set up data collection for Managed Service for Prometheus

The service offers both managed and self-deployed collectors. Get step-by-step instructions for setting up each option.

Query data from Managed Service for Prometheus

Query the data sent to the service using the Prometheus HTTP API, Prometheus UI, Grafana, the service page in the Google Cloud Console, and Cloud Monitoring.

Rule evaluation and metric filtering

Learn how to use functions you expect from Prometheus such as rule evaluation and metric filtering.

Use cases

Use cases

Use case
Diagnose problems with your applications quickly

Use PromQL to define alerts and diagnose issues when alerts are triggered. With Managed Service for Prometheus, you do not have to change your visualization tools or alerts so your existing incident creation and investigation workflows will continue working.  

Use case
Cost-effectively monitor dynamic environments

Managed Service for Prometheus charges on a per-sample basis, which does not charge for cardinality up front when a new container is spun up. With per-sample pricing, you only pay while the container is alive, so you are not penalized for using Horizontal Pod Autoscaling. Managed Service for Prometheus features other cost controls such as customizable sampling periods, filters, and the ability to keep data local and not send it to the datastore.

All features

All features

Stand-alone global rule evaluator You can continue to evaluate your existing recording and alerting rules against global data in Managed Service for Prometheus. The results are stored just like collected data, meaning you will not need to co-locate aggregated data on a single Prometheus server.
Dynamic multi-project monitoring Metrics scopes are a read-time-only construct in Cloud Monitoring that enables multi-project monitoring via a single Grafana data source. Each metric scope appears as a separate data source in Grafana and can be assigned read permissions on a per-service account basis.
Managed collectors Managed collectors are automatically deployed, scaled, sharded, configured, and maintained. Scraping and rules are configured via lightweight custom resources (CRs).
Self-deployed collectors Use your preferred deployment mechanism by simply swapping out your regular Prometheus binary for Managed Service for Prometheus’ collector binary. Scraping is configured via your preferred standard method and you scale and shard manually. Reuse your existing configs and run both regular Prometheus and Managed Service for Prometheus side by side.
Support for non-GKE and non-Kubernetes monitoring The managed collector can be used in non-GKE environments and guidance is provided for setup. As Prometheus itself can be configured to collect data from non-Kubernetes targets such as VMs, self-deployed collectors can be configured to collect data from VMs as well.
Backed by Monarch, Google’s in-memory time series database The service uses the same technology that Google uses to monitor its own services, meaning that even the largest Prometheus deployments can be monitored at global scale.
Cost control mechanisms Help keep your spending under control with an exported metrics filter, a reduced charge for sparse histograms, a fee structure that charges less for longer sampling periods, and the ability to only send locally pre-aggregated data.
Cost identification and attribution Use Cloud Monitoring to break out your Prometheus ingestion volume by metric name and namespace. Quickly identify the metrics that cost you the most and which namespace is sending them.



Pricing tiers are progressive, so usage will be charged at lower tier rates before being charged at higher tier rates. For example, if your total usage is 300 billion samples, your first 50 billion samples cost $0.15/million samples, the next 200 billion samples cost $0.12/million samples, and the final 50 billion samples cost $0.09/million samples.

Feature Price Free allotment per month Effective date
Metrics from Google Cloud Managed Service for Prometheus

$0.15/million samples: first 0-50 billion samples#

$0.12/million samples: next 50-250 billion samples

$0.09/million samples: next 250-500 billion samples

$0.06/million samples: >500 billion samples

Not applicable May 16, 2022
Monitoring API calls

$0.01/1,000 Read API calls

(Write API calls are free)

First 1 million Read API calls included per billing account July 1, 2018

 Google Cloud Managed Service for Prometheus uses Cloud Monitoring storage for externally created metric data and uses the Monitoring API to retrieve that data. Managed Service for Prometheus meters based on samples ingested instead of bytes to align with Prometheus' conventions. For more information about sample-based metering, see Pricing for controllability and predictability. For computational examples, see Pricing examples based on samples ingested.

#Samples are counted per billing account.