Code360 powered by Coding Ninjas X Code360 powered by Coding Ninjas X
Table of contents
How alerting works
How to add an alerting policy
How to manage to alert policies
Authorization required to create alerting policies
Required Google Cloud console roles
Required API permissions
Determining your role
Costs associated with alerting policies
Types of alerting policies 
Metric-absence condition
Metric-threshold condition
Frequently Asked Questions
What does alerting accomplish?
What distinguishes alerting from monitoring?
What is a tool for alert monitoring?
What do monitoring and metrics mean?
Last Updated: Mar 27, 2024

Introduction to alerting Part-1

Author Aditi
0 upvote
Master Python: Predicting weather forecasts
Ashwin Goyal
Product Manager @


Alerting lets you know when there are issues with your cloud applications so you can fix them right away. An alerting policy in Cloud Monitoring specifies the situations and methods for which you should be notified. The alerting guidelines are summarized on this page. Metric-based alerting policies keep track of the metrics data that Cloud Monitoring collects. Most of the Cloud Monitoring documentation on alerting policies uses metric-based alerting policies.

You can also establish log-based alerting policies that inform you when a specific message appears in your logs. These regulations do not rely on metrics. This article will discuss alerting, types of alerting policies, the behavior of metric-based alerting policies, and how to add severity levels to alerting policies.

Let’s dive into the article for more detail on alerting.

How alerting works

The following are the details for each alerting policy:

  • Conditions specify when one or more resources are in a situation that necessitates your response. Although you can set up a policy that contains several states, an alerting policy must have at least one condition.
  • As an illustration, you could set up a condition like this:
  • The HTTP response latency is higher than two seconds for at least five minutes.
  • In this illustration, the condition determines when you must answer based on the metric HTTP response delay values. The resource, or combination of resources, must be in a state that necessitates your response for the condition to be true.
  • Channels for notifications that specify who should be informed when action is necessary. An alerting policy may include several notification channels. Cloud Monitoring supports Cloud Mobile App and Pub/Sub along with standard notification channels. See Notification options for a complete list of supported channels and details on how to set them up.
  • For instance, you could set up an alerting policy to send an email to and post a message to the #my-support-team channel in Slack.
  • Information that you want to be mentioned in a notification. Variables, Markdown, and plain text are supported in the documentation field.
  • You may, for instance, incorporate the following material into your alerting policy:
## HTTP latency responses
This alert originated from the project ${project}, using
the variable $${project}

The conditions of a metric-based alerting policy are continuously monitored by Monitoring when it is configured. The conditions cannot be configured only to be observed during specific times.

Monitoring creates an incident and delivers a notification when the policy requirements are satisfied. An overview of the event, a link to the Policy details page so you may look into it, and any supporting material you supplied are all included in this message.

When a metric-based policy's requirements are no longer met while an incident is still active, Monitoring automatically closes the incident and notifies the user of the closure.


You deploy a web application on a Compute Engine virtual machine (VM) instance currently running a web application. Although you anticipate that the HTTP response latency will vary, you still want your support team to reply if the application experiences excessive latency for an extended period.

You develop the following alerting policy to make sure that your support team is informed when your application has excessive latencies:

Open an incident and contact your support staff via email if the HTTP response latency is more significant than two seconds for at least five minutes.

This alerting strategy requires keeping track of the HTTP response latency. The criteria is met, and an incident is produced if this delay is more than two seconds consistently for five minutes. The criterion is not satisfied, or a brief spike does not trigger an event in latency.

Due to increased demand, your web application's response time exceeds two seconds. Your alerting policy will react as follows:

  • When Monitoring gets an HTTP latency measurement greater than two seconds, a five-minute countdown is set in motion.
  • The timer runs out if the subsequent five latency measurements are more than two seconds apart. When the timer goes off, Monitoring declares the condition to have been satisfied, creates an incident, and emails your support staff.
  • The member of your support staff who received the email logs into the Google Cloud dashboard confirmed that they had received the message.
  • Your support staff can address the reason for the lag by using the documentation in the notification email. The HTTP response latency is less than two seconds after a few minutes.
  • Monitoring resolves an incident and notifies your support team that it has been closed when it receives an HTTP latency measurement of fewer than two seconds.

A new incident is opened, and a notification is delivered if the latency exceeds two seconds and continues to exceed that level for five minutes.

How to add an alerting policy

Using the Google Cloud dashboard, the Cloud Monitoring API, or the Google Cloud CLI, you may create a metric-based alerting strategy for your Google Cloud project:

  • When using the Google Cloud console, you can enable a suggested alarm or create your alert by beginning on the Alerts page of Cloud Monitoring.
  • There are accessible recommended notifications for several Google Cloud products. The only configuration necessary for these alerts is the addition of notification channels. For instance, the Pub/Sub Lite Topics page contains links to alerts set up to send you an email when your subscription quota is about to be reached. Similar to this, the VM Instances tab in Monitoring provides links to alerting policies that are set up to track those instances' network latency and memory usage.
  • Using the Cloud Monitoring API or the Google Cloud console, you can examine and edit any policy that you establish with the Google Cloud console. The Cloud Monitoring API enables you to develop alerting strategies that keep track of metric ratios. You cannot access or amend these policies when they employ Monitoring filters using the Google Cloud panel.
  • You can make, view, and edit alerting policies directly using the Google Cloud CLI or the Cloud Monitoring API. Using the Google Cloud CLI or the Cloud Monitoring API, you may define conditions that keep an eye on a ratio of metrics. Utilizing Monitoring Query Language (MQL) or Monitoring filters, you can provide the ratio using the Cloud Monitoring API. See Metric ratio for an illustration of a policy that employs Monitoring filters.

The Google Cloud console and the Cloud Monitoring API enable an expressive, text-based language that Cloud Monitoring supports. Create Alerting Policies Utilizing Monitoring Query Language (MQL) for details on using this language with alerts.

Use the Logs Explorer in Cloud Logging or the Monitoring API to implement a log-based alerting strategy for your Google Cloud project. See Monitoring your logs for details on log-based alerting policies.

How to manage to alert policies

See the following for instructions on how to access a list of your project's metric-based alerting policies and how to change those policies:

  • Making use of the Google Cloud console to manage to alert policies
  • Using the Cloud Monitoring API or Google Cloud CLI to manage to alert policies

See Using log-based alerts for more on managing log-based alerting policies.

Authorization required to create alerting policies

The roles or permissions needed to create an alerting policy are described in this section. Access control has further information on Identity and Access Management (IAM) for Cloud Monitoring.

Each IAM role is assigned a name and an ID. When defining access control, role IDs are supplied as parameters to the Google Cloud CLI and take the form of roles/monitoring.editor. See Granting, modifying, and canceling access for additional details. Names of roles, including Monitoring Editor, are displayed in the Google Cloud console.

Required Google Cloud console roles

Your IAM role name for the Google Cloud project must be one of the following to build an alerting policy:

  • Monitoring Editor
  • Monitoring Admin
  • Project Owner

Required API permissions

Your IAM role ID for the Google Cloud project must be one of the following to utilize the Cloud Monitoring API to construct an alerting policy:

  • roles/monitoring.alertPolicyEditor: The basic minimum permissions required to develop an alerting policy are provided by this role ID.
  • roles/monitoring.editor
  • roles/monitoring.admin
  • roles/owner

Determining your role

Use the Google Cloud console to identify your project role by performing the following actions:

  • Select the Google Cloud project after launching the Google Cloud console:
  • Go to the Google Cloud console
  • You can click IAM & admin to see your role. On the same line as your username is your role.

Contact the administrator of your organization to learn more about your organization-level permissions.

Costs associated with alerting policies

Use of alerting policies or uptime checks is free, although the following restrictions do apply:

Costs associated with alerting policies
Get the tech career you deserve, faster!
Connect with our expert counsellors to understand how to hack your way to success
User rating 4.7/5
1:1 doubt support
95% placement record
Akash Pal
Senior Software Engineer
326% Hike After Job Bootcamp
Himanshu Gusain
Programmer Analyst
32 LPA After Job Bootcamp
After Job

Types of alerting policies 

This document includes JSON samples for several metric-based alerting strategies and describes them in detail. Alerting policies outline three conditions to look out for: Some metrics act in a certain way over a certain amount of time. For instance, when a metric's value exceeds a certain threshold or when it fluctuates too quickly, an alerting policy might be triggered.

Metric-absence condition

A metric-absence condition is triggered when there are no values for a particular time window in a monitored time series.

Note: Metrics associated with TERMINATED or DELETED Google Cloud resources are not considered for metric-absence policies, except for metrics produced by an uptime check. Using metric-absence policies, you cannot test for terminated or deleted Google Cloud VMs. Additionally, uptime generates check-results metrics while running whether an uptime check succeeds or fails. As an illustration, this means metrics like shouldn't be alerted to by metric-absence.

Metric-absence conditions demand at least one productive measurement or one that retrieves data within the maximum time frame following the installation or modification of the policy. If you use the Google Cloud console, the maximum adjustable duration window is 24 hours; if you use the Cloud Monitoring API, it is 24.5 hours.

Consider the scenario where the duration window for a metric-absence policy is set to 30 minutes. When the component that writes metric data has never written a data point, the condition won't occur. The subsystem must produce at least one data point before ceasing all data production for 30 minutes.

Metric-threshold condition

When a metric's values for a given duration window are more significant than or less than the threshold, a metric-threshold condition is triggered. For instance, a metric-threshold condition might be fulfilled when the CPU utilization is greater than 80% for at least 5 minutes.

Patterns that fit into one of the following main sub-categories are found within the class of metric-threshold conditions:

  • When values in a time series increase or drop by a specified percentage or greater throughout a period window, rate-of-change requirements are met.

This condition computes the percent-of-change on the time series before comparing it to the threshold.

The condition averages the metric's values for the previous 10 minutes, then compares the result to the average for the 10 minutes taken immediately before the duration window. A metric rate of change condition uses a 10-minute lookback window, a fixed value that cannot be altered. However, when establishing a condition, you define the time window.

  • Group-aggregate conditions are triggered when a metric calculated across a resource group exceeds a threshold for a duration window.
  • When an uptime check cannot correctly respond to a request from at least two different locations, uptime-check conditions are set off.
  • Process-health conditions are triggered when the number of processes executing on a VM instance is greater than or fewer than a threshold. These conditions can also be set up to monitor a collection of instances with similar names. The Ops Agent or the Monitoring Agent must be running on the monitored resources for this condition type to exist.
  • When the ratio of two metrics exceeds a threshold for a duration frame, metric-ratio conditions are triggered. These conditions compute the ratio of two metrics, such as the proportion of HTTP error responses to all HTTP responses.

Note: The Google Cloud console includes all alerting rules, but unless the policies employ Monitoring Query Language, you must use the Cloud Monitoring API or the Google Cloud CLI to create, view, or amend ratio-based policies. 

Frequently Asked Questions

What does alerting accomplish?

People may stay updated on the information that matters most thanks to alerts. The most prevalent usage of the service is for machine-to-person communication, with alerts often given through a notification system.

What distinguishes alerting from monitoring?

The most straightforward kind of monitoring is the display of metrics through dashboards and reports. Alerts need action, such as resuming a service, notifying a person, updating a log, etc.

What is a tool for alert monitoring?

IT professionals can monitor servers to track information about their performance, health, and status, including CPU load, memory usage, active processes, and disc space levels.

What do monitoring and metrics mean?

Software metrics quantify a software's properties in a quantifiable way. Because of this, monitoring the metrics plays a significant role in the development phase. During the development and deployment phases, monitoring system metrics aim to ascertain the product's or process's quality.


In this article, we have extensively discussed the introduction of alerting in GCP. We have also explained how alerting works, authorization required to create alerting policies, the types of alerting policies, and other details.

We hope this blog has helped you enhance your knowledge about alerting. If you would like to learn more, check out our articles on introduction to cloud computingcloud computing technologiesall about GCP and AWS Vs. Azure Vs. Google Cloud. Practice makes a man perfect. To practice and improve yourself in the interview, you can check out Top 100 SQL problemsInterview experienceCoding interview questions, and the Ultimate guide path for interviews.

Do upvote our blog to help other ninjas grow. Happy Coding!

thank you image
Previous article
Overview of uptime checks
Next article
Introduction to alerting Part-2
Live masterclass