GraphQL Metrics

With no configuration, get real-time observability for your GraphQL APIs usage, performance and errors.

After every release, we open up Stellate's GraphQL Metrics to monitor the health of our GraphQL API in real-time.
Charles Brown's avatar

Charles Brown

Head of Application Services, Formidable Labs

Metrics features showcase
Analytics Features

Knowledge is performance

  • Get real-time insights into all GraphQL request

    Get visibility into every single query and mutation sent to your GraphQL API, including the request’s headers, latency, errors, and more.

    On the left we see a chart displaying 4 stats, a blue line for the cache hits, a red line for the cache misses, a grey line for the cache passes and an orange bar for the amount of errors. On the right we see these stats written out, 86.4 million requests, which is a 4.1% improvement, of those requests 44.4 million are cache-hits, 11.5 million are cache-misses and 30.4 million are cache-passes and we have seen 1.08million errors.

  • Monitor performance, identify bottlenecks

    Automatically oversee the performance of your GraphQL API down to the individual operation level to identify your biggest bottlenecks.

    a query named getProductWidgets is displayed accompanied by a P99 latency of 236ms, it tells us that with Stellate Edge Cache enabled it improves speed by 404ms and the query has seen 6.18million requests.

  • Track HTTP and GraphQL errors

    Track all HTTP and GraphQL errors and receive alerts on spikes the moment they become a problem for your users.

    On the left a bar chart showing the amount of errors we have seen, on the right we see a total of 1.09 million errors have happened, which is a 98% increase, of these errors we have seen 1.06million GraphQL errors, 570 thousand HTTP 4xx Errors and 29869 HTTP 5xx Errors.

  • Analyze issues quickly with pinpoint precision

    Deep dive into your data at minute-level accuracy, aid your debugging process, and inform decision-making.

    A date picker showing the date range from the 9th of March 2023 to the 15th of March 2023

  • Unlock the Potential of Caching

    Analyze operations cardinality to identify areas for caching improvements to boost Cache Hit Rate across all operations.

    A card showing the Cache hit rate, Max CHR, Total number or requests, Distinct IPs, Distinct variable combinations, Distinct header combinations, Distinct response combinations, Other distinct request parts

  • Understand Your Users Engagement

    Track data points with each GraphQL request with Custom Attributes. Make your metrics multi-tenant and enhance user journey debugging. Your API insights, personalized.

    A card showing two Custom Attributes, one called "company" showing the top three companies that send the most requsts, the other one called "locale" showing three most used languages

  • Dive deep with fine-grained filtering

    Filter your metrics down to the nuances. Check the performance of certain operations or discover how certain HTTP referrers cause errors.

    The applied filters, this highlights that we are filtering by country which has to be Belgium or the United Kingdom, the HTTP Refferer which has to be acme.com, the browser which has to be Chrome and finally the types which have to be AdminAPIToken or Analytics or Alert.

  • Get real-time alerts and manage incidents

    Notify the right people over the right channel when things go wrong. Use the Slack and PagerDuty integrations or good old email.

    A slack message telling us that an app named Stellate has sent a message alerting us of an increase in the amount of 5xx HTTP errors, we have seen 85 of them in the last minute. Then the message shows us the last error, which is a message telling us that we can't reach the endpoint, lastly we can click a button to go look at all the errors

  • Prevent breaking your GraphQL schema

    Identify unused fields and types and prevent breaking changes to your schema based on actual usage data from your users.

    A terminal message telling us that we have made a breaking change to the "inviteUser" mutation by changing the return type from Invite to Organization
Combine with caching

Works great with Stellate GraphQL Edge Caching

  • Understand your global latency

    Most GraphQL APIs sit in one data center. Monitor your global latency and verify the performance boost your users get through edge caching.

    Map showing P95 latencies for North America (512ms), South America (1208ms), Europe (159ms) and Africa (747ms)

  • Get cache hit rate insights

    Understand how the attributes of your various data types affect the cache hit rate of individual queries.

    A card where we see Different types of our application with their corresponding cache hit rate on the left side, on the right side there are different unused types and when they were last seen

  • Monitor how your Cache Hit Rate changes

    Understand your caching strategy by observing Cache Hit Rate over time. Analyze trends, identify patterns, and adapt your cache rules to boost your API caching performance.

    A card where we see the Cache Hit Rate chart over a period of time

  • Optimize your cache rules performance

    Evaluate cache rules performance. Optimize your caching strategy by understanding how requests and operations are affected by your cache rules.

    Two cache rules showing the details and the affected types

  • Check your purging analytics

    Whether auto invalidated from mutations or manually invalidated from the API, Metrics show when and what is being purged from the cache.

    A statistic showing that a mutation named updateOrder resulted in an expected latency increase of 1537ms because it invalidated a set of types and queries

  • Become a pro in cache debugging

    Track every request and its cache state. Monitor your cache hit rate and understand why certain requests are being cached and others are not.

    A query named getCategories resulted in a cache-pass because the request body was too big, we can also derive that we retried the request three times, it it failed twice due to a Bad Gateway error and ended up in a success.

  • One-Click Cache Purging

    Stay in control of content changes even if you can't rely on webhooks. Instantly purge the entire cache or use Purge actions to remove cache items based on specific operations or types from our metrics dashboard.

    A query named caseStudyCollection is displayed accompanied by a P99 latency of 103ms alongside a Purge button that allows you to purge the cache for this query

What are you waiting for?

It takes less than 5 minutes to set up a fully functional CDN for your GraphQL API.