Axiom and Serverless: The perfect pairing for o11y without limits

Takeaways:

The evolution of serverless observability

Since its release in 2015, the Serverless Framework has emerged as the go-to solution for building and deploying serverless applications across multiple cloud providers. With more than 300 million downloads and 46 thousand GitHub stars, it’s proven the open-source option of choice to simplify the development and management of event-driven, pay-per-execution compute services.

As enterprises seek to optimize resource utilization and accelerate time to market, serverless architectures have become the norm. According to a 2023 report, 70% of AWS users now leverage its serverless Lambda functions, with similar growth across serverless functions in Azure and Google Cloud. The Serverless Framework capitalizes on this trend by offering a unified, provider-agnostic approach to serverless development that enables organizations to avoid vendor lock-in while streamlining their cloud operations.

As digital transformation initiatives accelerate across industries, serverless architectures address several key enterprise concerns: cost efficiency, scalability, and developer productivity. The Serverless Framework, used by more than 90% of organizations running AWS CloudFormation, enhances these benefits by providing a consistent development experience, a robust plugin ecosystem, and built-in best practices for both security and performance optimization.

CloudWatch will do … until it won’t

AWS has established itself as a top choice for serverless developers, largely due to its mature and well-integrated serverless components: Lambda, DynamoDB, and S3.

CloudWatch, AWS’s monitoring and observability service, plays a crucial role in managing serverless applications. CloudWatch Logs Insights in particular serves as a comfortable starting point for observability needs on AWS because of two turnkey features:

• Native integration: CloudWatch Logs is tightly integrated with AWS services, especially Lambda. By default, Lambda functions send their logs to CloudWatch Logs, providing immediate visibility into function executions without additional configuration.
• Log insights: The CloudWatch Logs Insights feature provides a query language for analyzing log data, to help developers gain deeper insights into application behavior and performance.

CloudWatch sounds great — and is, at first. But as businesses grow and developers’ needs grow, too, many find themselves running into limits on what they can do with CloudWatch:

• Costs: as serverless usage grows, CloudWatch can shoot to the top of monthly spending. Unexpected, eye-watering bills have hit many a team out of the blue.
• Sources outside AWS: CloudWatch Logs Insights’ AWS-centric nature can require extra effort and tooling to get a unified view across different cloud providers or on-prem systems.
• Limited query capability: Its two dozen or so operators and functions let developers fall far short of the advanced queries necessary for insight into more complex, nuanced or irregular issues.

Axiom’s integration with Serverless

As applications grow in complexity, developers seek more advanced observability tools that offer distributed tracing, advanced analytics, and enhanced visualization capabilities. Axiom and Serverless are a perfect match — both Axiom and Serverless are developer-first companies focused on DX and architected from the ground up to optimize and maximize utility and value, not only on AWS but in multi-cloud environments that are only growing in size and scope.

Severless’s announcement of the integration says, “Axiom provides a cost-effective, powerful and elegant alternative without compromising on essential features or performance.”

How elegant is the Axiom / Serverless integration? It’s one line of config in your serverless.yml file. Yet Axiom provides developers with a featureful answer to their observability needs:

• Axiom slashes costs to a fraction of CloudWatch Logs. Many customers report more than 70% savings.
• It’s easy to aggregate logs from anywhere, inside or outside AWS and Serverless.
• Axiom’s APL query and processing language goes far, far beyond not only CloudWatch but most other alternatives.

If you’d rather get started than keep reading, see Serverless’s documentation for setup instructions.

Why Axiom is the best choice

Axiom was founded on a vision that the limits of log management could be obliterated by a fully cloud-based architecture that optimized every step of log management to make the most of still-evolving cloud environments: Schema-free ingest pipes with smart routing for durability at petabyte scale. A storage layer that adapts to patterns in event streams to continually maximize compression of incoming data. A serverless query engine whose design makes resource contention between queries, users or teams impossible.

• Axiom makes it affordable and performant to capture and keep 100% of your events. No sampling, no filtering, no cold storage. You can instantly query your entire event history whenever you want.
• Axiom can easily slash CloudWatch spending by 70%. Here’s a cost comparison fo 5 TB per day of logs from function invocations, database operations, S3 access and other typical logs in a serverless AWS environment.

Price comparison for 5 TB of log data per day.

• Axiom’s pricing is not only low, it’s simple and predictable — especially when compared to Amazon’s. There are no surprise bills, and no need to stop and research a new idea’s potential add-on fees. There aren’t any. Instead of molding their thinking to fit their tools and budgets, Axiom customers unleash their creativity.
• Retention in Axiom is 95 days by default, compared to just 30 days in CloudWatch. Axiom offers the option to extend retention up to several years, compared to a maximum of 15 months in CloudWatch before you’ll need to create a system to store them yourself.
• Axiom can use open-source collectors such as FluentBit or Logstash to collect logs from many AWS services without routing them through CloudWatch. For vended logs (Amazon VPC, Route 53, VPC Flow, WAF Access and others), which are hardwired to go straight into CloudWatch, Axiom CloudWatch Forwarder can extract them immediately into Axiom. Their retention time in CloudWatch can then be minimized to remove that cost.

Beyond basics: APL’s power in action

While CloudWatch Logs Insights offers a starting point, APL takes serverless observability to a whole other level for hundreds of AWS customers. These snippets showcase some of APL’s rich capabilities:

// Dynamic filtering
| where iff(lambda_name_filter != "*", ['lambda.name'] == lambda_name_filter, true)

// Regex-powered error detection
| where message matches regex "^[0-9.:TZ-]{20,}\\s+[0-9a-f-]{36}\\s+ERROR\\s+"

// Multi-dimensional analysis
| summarize Duration = max(durationMs) by bin_auto(_time), ['Lambda Name']

// Comprehensive cold start analysis
| extend initDurationMs = column_ifexists("record.metrics.initDurationMs", real(null))
| where type == "platform.report" and isnotnull(initDurationMs)
| summarize ['Cold Starts'] = count() by bin_auto(_time), ['Lambda Name']

• Dynamic filtering: Easily create flexible dashboards that adapt to your needs.
• Advanced pattern matching: Catch elusive errors that simple keyword searches miss.
• Multi-dimensional analysis: Slice and dice performance data any way you need.
• Granular cold start tracking: Analyze cold starts across time and functions for targeted optimization.

But Axiom’s real superpower? Lightning-fast queries on massive datasets. Unlike CloudWatch, which crawls on large volumes, Axiom maintains blazing speed whether you’re analyzing a day or a year of data.

Axiom vs CloudWatch: More power, less friction

Beyond basics: Advanced observability

Axiom enables deeper insights into serverless event data. To start with, it automatically creates an AWS Lambda dashboard if it detects Lambda events:

Beyond that, it’s easy and quick to create dashboards to monitor and improve performance of serverless apps. Here’s one to help optimize slow-performing Lambda queries, grouping logs with Lambda Invocations and Execution Time by function:

Axiom Lambda function monitors let you identify the different points of invocation failures, cold starts delays, and AWS Lambda errors on your Lambda function, alerting you to significant spikes:

Axiom also has Smart Filters. These let you easily filter down to specific AWS Lambda functions or Serverless projects, and use our saved queries to get deep insights on how functions are performing with a single click:

See beyond Serverless and AWS

Axiom connects to the larger ecosystem outside Serverless and AWS to extend its coverage to multi-cloud environments. It fully supports OpenTelemetry for tracing, and is widely used with modern collectors like Vector, which supports more than 40 sources.

Customers report unexpected benefits

All of this isn’t a hypothetical use case — it’s been widely deployed for years. Customers love Axiom as a CloudWatch replacement, often for reasons they’ve discovered on their own:

“Going from CloudWatch to Axiom feels very natural. With APL for queries, it’s the same muscle memory, the same workflow.”

Andres Hernandez

Co-founder and CTO, Hapn

“Axiom feels much more like a building block of your application rather than an internal tool.”

Ryan Fox-Tyler

SVP of Products and Engineering, Hypermode

“We set up Axiom so we could troubleshoot our customers, and that evolved into, ‘What if they didn't have to ask us about it?’”

Shawn Rushefsky

Senior Generative AI Solutions Architect, Salad

“If you have a lot of log volume, I would say go to Axiom.”

Ray Jacobson

VP of Engineering, Audius

The future: Event logs as organizational intelligence

Axiom and Serverless agree that as AI improves rapidly and gains footing inside organizations, the logs used today for monitoring and o11y will become foundational intelligence of behavior both inside and outside the organization. A core reason Axiom was founded with the vision of ingesting, storing and querying 100% of event data is that companies will want to train burgeoning AI tools on every bit of information possible. Many customers have already begun building organatizonal insight — e.g. a business health dashboard used by the CEO — atop Axiom, and plan to make it a foundational component of their AI-driven future.

Sampling events and cherry-picking sources isn’t a best practice. It evolved as a coping mechanism when log volumes and the associated costs outgrew older log management systems designed with far fewer events in mind. Engineers only throw away data for one reason: to save money. Axiom was designed to remove that need. Its integration with The Serverless Framework makes it both affordable and easy for fast-growing organizations, in which event records are becoming foundational knowledge, to capture and keep it all.

Start saving today, build for tomorrow

To set up the Axiom integration, see Send data from Serverless to Axiom in the documentation. To talk to us about enterprise-scale evaluation and integration, email sales@axiom.co.