Topics

Audit Logs, Part 2: A Cross-Service Comparison of Web Service Logging

  • column

What Remains in AWS, Cloudflare, Vercel, Supabase, and Headless CMS?

Hello, I'm Otsuka, CTO at Liberogic.

Last time, we outlined the differences between access logs, application logs, and audit logs.

This time, we take a more concrete look at the kinds of logs you can get from AWS, Cloudflare, Vercel, Supabase, and headless CMS platforms, which we work with regularly.

That said, this is not a detailed comparison of each service's pricing or features.

Instead, we give a brief overview of how the services differ considerably in terms of "what is retained by default" and "how far back you can review what happened."

Services

Logs You Can Typically Review

Approach to Long-Term Retention

Points to Note

AWS

Execution, access, and management operations

Combine with CloudWatch or S3

Requires configuration for each service

Cloudflare

Worker execution, exceptions, and more

Forwarded to R2 or external services

Static delivery is separate from execution logs

Vercel

Functions, deployments, and execution logs

Forwarded externally via Log Drains and similar

Retention period depends on your plan

Supabase

Database, API, authentication, and admin operations

Forwarded externally via Log Drains and similar

App-specific operations are implemented separately

Headless CMS

Content updates and admin operations

Uses the service's history and audit features

Target operations and plans differ

Datadog/Sentry

Aggregated logs and error information

Designed according to purpose and storage volume

Adding the feature alone does not create an audit log

Even with the same "logging feature," the operations covered and retention periods can differ.

Also, being visible in the admin screen and being retained long-term for auditing are separate matters.

AWS offers a full set of logging tools, but configuration is still required.

AWS offers a particularly rich set of services for logging.

For example, execution status for services such as Lambda and API Gateway is collected in CloudWatch Logs. Operations performed in the AWS console or through the API can be recorded with CloudTrail.

Put simply,

  • CloudWatch Logs: Monitor the behavior of applications and AWS services
  • CloudTrail: Track who did what in your AWS environment
  • S3: Store logs for the long term

This is how the roles are divided.

Search, monitoring, alerting, and long-term storage can all be built within AWS, which makes it easier to support systems with strict audit requirements.

However, using AWS does not mean that everything is automatically logged.

You need to enable logging for each service you use and decide on retention periods and storage locations. Because the flexibility is high, the design must be done carefully.

Separate short-term debugging from external log storage with Cloudflare

In Cloudflare Workers, you can check output from console.log() and errors in Workers Logs.

Workers Logs are very useful for development checks and investigating recent incidents. However, if you need to keep a long-term audit trail, you should not rely solely on the standard log dashboard.

For logs that must be retained long term, consider an architecture that sends them to R2 or an external log management service using Logpush or similar tools.

Also, if you serve a static website on Cloudflare Pages or Workers, not every request is recorded in the Worker execution logs.

Since the site displays normally, it's easy to assume that access logs must exist. However, serving static files and running Workers are separate processes.

Cloudflare makes it relatively easy to build compact architectures, but you need to understand which processes handle which requests in order to design logging effectively.

Vercel Makes Next.js Logs Easy to Check

On Vercel, you can view logs output by Next.js Functions, Server Actions, and other features from the dashboard.

Deployments and runtime logs are located close together, making the environment easy for developers to understand and convenient for everyday error investigation.

That said, how long logs can be retained and the scope in which they can be forwarded externally depend on your plan.

If you want to retain logs for a longer period, you will need a separate mechanism, such as using Drains to send them to an external storage destination.

Regardless of Vercel, having logs visible in the dashboard and having them stored long-term for audit purposes are two different things.

It is important to check these two separately.

Supabase has many logs beyond the database.

Supabase is centered on PostgreSQL, but in practice it is made up of multiple features beyond the database, including authentication, APIs, storage, and Edge Functions.

Since you can check each log from the dashboard, it is easy to follow how the application as a whole behaves.

There are also authentication logs, as well as audit logs for checking actions performed in the Supabase dashboard.

Note, however, that Supabase administrative actions and operations within your own application are separate.

For example, the service can record who changed a Supabase project's settings, but to record who changed customer information in your own admin screen, you need to implement logging on the application side.

Also check the operation history of your headless CMS

Headless CMSs such as microCMS, Kuroco, and NILTO provide content update history and operation logs for their admin panels.

Records showing who updated an article, when it was published, and when settings or permissions were changed are important as audit logs for website operations.

However, the actions recorded, the retention period, and the plans that include these features vary by service.

Even if you can view content update history, not every administrative action is necessarily kept as an audit log.

Also, a CMS basically only holds records of activity within the CMS itself. Access to the published website and errors occurring on the frontend need to be checked in the hosting environment.

In headless CMS projects, it is easier to understand the system by viewing the CMS, frontend, hosting, and API as one system rather than as separate parts.

When adding Datadog or Sentry

When logs are spread across multiple services, every incident requires opening each admin panel separately to investigate.

Consolidating logs into a log management service such as Datadog lets you search logs from multiple environments together and detect anomalies and send notifications.

Sentry is also widely used, but it specializes in identifying where errors occur and understanding their impact.

Both are useful, but simply adopting them does not automatically create a complete audit log.

You need to decide in your application what to output, and storing all logs over the long term increases costs as data volume grows.

Storage method

Best suited for

Features

Admin console and log management services

Routine failure investigation

Logs can be searched quickly, but costs increase with storage volume

Storage such as R2 and S3

Audit and long-term log retention

Low-cost to keep, but retrieval takes extra effort during an investigation

A practical approach is to separate logs you search regularly from logs you keep on hand in case something goes wrong.

You can't decide which service is better

You can't simply compare logging features and declare AWS or Cloudflare the winner.

What matters is what you need to record for that specific project.

  • I want to investigate service errors
  • I want to record administrator actions
  • I want to review CMS update history
  • I want to track authentication and permission changes
  • We want to retain it for a long period for auditing purposes

Once you know what you need, you can see which parts are covered by each service's standard features and which parts require additional mechanisms.

We also don't build a large logging foundation from the start. Instead, we design the setup according to each project's requirements and budget.

Next time, we'll take a closer look at practical details: when these services are combined, what kind of logs the application should output and where they should be stored.

Well then.

About the author

The backbone of Liberogic's technical division. Whenever he hears someone say, "I wish we had something like this, it would be so handy," he adds his own ingenuity, turns the idea into added value, and gets it implemented in no time. A highly sociable team member who has many fans among our clients, he is a treasured asset to our company and a devoted cat lover.

Sho Otsuka

Director, CTO / Chief Engineer / Representative, Nekoana LLC / Looks unreasonably young

View this staff member's articles

Reliable team structure and responsive support are our strengths

Liberogic's experienced staff actively drive projects forward, which has earned high praise from our clients.
We make sure to properly assign project managers and directors to keep the entire project running smoothly. Our style of allocating resources to the right people, rather than adding full-time commitment unnecessarily, helps prevent cost increases. We are also known for our speed, from understanding the scope of work to preparing and submitting estimates.

Note: Please be advised that we do not actively engage in SES-style on-site staffing assignments.

We support nearly all major project management and chat tools, including Slack, Teams, Redmine, Backlog, Asana, Jira, Notion, Google Workspace, Zoom, and Webex.

Have a web challenge? Let us know.

Case Studies