@mihilmy/cw-logger
v1.0.6
Published
Aggregated cloudwatch logs publisher
Downloads
10
Maintainers
mihilmyReadme
Library allows users to aggregate metrics and raw logs using the Embedded Metric Format, saving costs and providing raw data for analysis. Designed to be used on your frontend applications where each session is tracked with its own log stream, allowing high throughput and high cardinality metrics per user session.
📖 Concepts
Reporter: Responsible for reporting logs to CloudWatch every X seconds, partitioning into multiple requests if necessary and handling a retry strategy for throttles.Aggregator: Collects metrics/logs throughout the session by acting as an in-memory storage layer, these are later sent to cloudwatch via aReporter.
🔌 Extend It
Library was designed carefully to allow users to extend the basic usage by adding a more complex aggregation logic to optimize for costs,
you can implement the Aggregator interface, and the library should continue to work as expected. You have the ability to create custom
metric classes and factories.
🔐 Authorization
Request and response authorization is beyond the basic reporter capability. It's recommended having an endpoint in your backend that can proxy these requests to the cloudwatch endpoint. The most ideal design is when using it with a CloudFront distribution, and you can hookup a request signer that passes the requests to the cloudwatch origin.
🌏 Web Applications
// Step 1: Initializer a reporter instance
const reporter = new WebSessionMetricsReporter<SpotifyPlayerMetrics>({
namespace: "WebsiteMetrics",
aggregator: new BasicAggregator(),
logsEndpoint,
logGroupName,
flushFrequency
});
reporter.startReporting("@mihilmy");
// Step 2: Initialize handler for recording the user action
document.getElementById("save").onclick = () => reporter.addCount("SaveButtonClicks", 1);
<!-- Sample markup for a simple page -->
<body>
<div id="app">
<input type="text" id="Address" placeholder="Add Address" />
<button id="save">Save</button>
</div>
<script src="src/index.ts"></script>
</body>ƛ AWS Lambda
AWS Lambda makes it too easy to include custom metrics in your application, simply log the embedded metric format to your cloud watch logs and this will generate a metric in cloudwatch backend.
// Add this to a shared script that multiple consumers can use
const reporter = new LambdaMetricsReporter<BackendAPIMetrics>({ namespace: "BackendMetrics" });
reporter.addCount("XApiInvocations", 1); // One liner!!CloudWatch Hard Limits
Adding these here as a reminder for how much we can optimize
- The embedded metric format is subject to a maximum size of 256 KB.
- The batch log events are limited to 1MB and 10,000 events.
- Can not be more than 2 hours in the future.
- Can not be older than 14 days.
- Can not span more than 24 hours.
- The log events in the batch must be in chronological order by their timestamp
- 5 TPS hard limit on sending logs to a log group.