mikrometric

MikroMetric is a Lambda-oriented lightweight wrapper for producing AWS CloudWatch Embedded Metric Format-compatible metric logs.

---

MikroMetric is the quickest way to get going with AWS metrics (CloudWatch Embedded Metric Format):

• Adapted out-of-the box for serverless Lambda environments
• Zero config besides setting a namespace and service name
• Familiar syntax using methods like putDimension(), putMetric(), setProperty(), and flush()
• Easier to understand than raw EMF and less chance of breaking it (or your computer, trying to get it right!)
• Less bloated than the other major libraries
• Cuts out all the stuff you won't need in cloud/serverless
• Tiny (~2.2 KB gzipped)
• Has only one dependency, aws-metadata-utils (for picking out metadata)
• Has 100% test coverage

Usage

Basic importing and usage

// ES5 format
const { MikroMetric } = require('mikrometric');
// ES6 format
import { MikroMetric } from 'mikrometric';

// Minimal usage
const mikroMetric = MikroMetric.start({ namespace: 'MyNamespace', serviceName: 'MyServiceName' });

// Using the AWS `event` and `context` objects
const mikroMetric = MikroMetric.start({
  namespace: 'MyNamespace',
  serviceName: 'MyServiceName',
  event,
  context
});

mikroMetric.putDimension('user', 'Sam Person');
mikroMetric.putMetric('duration', 83, 'Milliseconds');
mikroMetric.setProperty('correlationId', '8d5a0ba6-05e0-4c9b-bc7c-9164ea1bdedd');

mikroMetric.flush();
Copy

Provided full event and context objects, your metric log will look something like this in CloudWatch Logs:

{
  "_aws": {
    "Timestamp": 1668191447669,
    "CloudWatchMetrics": [
      { "Namespace": "MyNamespace", "Dimensions": [["service"]], "Metrics": [] }
    ]
  },
  "accountId": "123412341234",
  "correlationId": "6c933bd2-9535-45a8-b09c-84d00b4f50cc",
  "functionMemorySize": "1024",
  "functionName": "somestack-FunctionName",
  "functionVersion": "$LATEST",
  "id": "3be48135-927a-4f71-ac7e-94bcd30723ba",
  "region": "eu-north-1",
  "resource": "/functionName",
  "runtime": "AWS_Lambda_nodejs16.x",
  "service": "MyService",
  "stage": "shared",
  "timestamp": "2022-11-11T18:30:47.669Z",
  "timestampEpoch": "1668191447669",
  "timestampRequest": "1657389598171",
  "user": "some user",
  "viewerCountry": "SE"
}
Copy

MikroMetric will grab as many values as it can from the event and context objects. You will of course get fewer (or none) of the dynamic metadata values if these objects are not provided.

The namespace and serviceName may be passed in either manually at init-time (as above), or be inferred via environment variables (see below). When initializing, some representation of these values must exist or an error will be thrown.

You can now use CloudWatch Logs Insights and CloudWatch Metrics to either search your logs or visualize your metrics.

For more learning resources regarding AWS observability solutions, see the One Observability Workshop, especially the page on EMF.

Setting custom static metadata

import { MikroMetric } from 'mikrometric';

// Use custom metadata
const metadataConfig = {
  version: 1,
  hostPlatform: 'aws',
  owner: 'MyCompany',
  domain: 'MyDomain',
  system: 'MySystem',
  team: 'MyTeam',
  tags: ['backend', 'typescript', 'api', 'serverless', 'my-service'],
  dataSensitivity: 'proprietary'
};

const mikroMetric = MikroMetric.start({
  namespace: 'MyNamespace',
  serviceName: 'MyServiceName',
  metadataConfig
});
Copy

Creating metric logs

Put a dimension

mikroMetric.putDimension('user', 'Sam Person');
Copy

Put a metric

mikroMetric.putMetric('duration', 83, 'Milliseconds');
Copy

Set a property

mikroMetric.setProperty('correlationId', '8d5a0ba6-05e0-4c9b-bc7c-9164ea1bdedd');
Copy

Flush

"Flushing", or sending, logs is easy:

mikroMetric.flush();
Copy

Because this is assumed to be within an AWS Lambda context, then flushing is simply a case of logging out the metric data.

Flushing will reset the MikroMetric instance. See Reset below.

Other features

Set the correlation ID manually

mikroMetric.setCorrelationId('abc123');
Copy

Set the namespace manually

mikroMetric.setNamespace('MyNewNamespace');
Copy

Get namespace

mikroMetric.getNamespace();
Copy

Get service name

mikroMetric.getServiceName();
Copy

Reset

This resets the metric log context and therefore erases any dimensions, metrics, and properties.

mikroMetric.reset();
Copy

Configuration

Setting the namespace and/or service name with environment variables

You can set MIKROMETRIC_NAMESPACE and/or MIKROMETRIC_SERVICE_NAME respectively as environment variables to use these at init time.

Any values manually passed in will always take precedence.

Behavior of output

Output fields will be spread (and potentially deduplicating same-named properties) in the following order:

1. Dynamic metadata (AWS)
2. Custom static metadata
3. Properties

A higher number means that fields in that category will persist if any fields have same names in lower-numbered categories.

License

MIT. See LICENSE file.