dreamorosi
commented
1 year ago As part of this issue I was thinking of addressing a couple of points: a. add the Idempotency utility to the examples b. refactor the part of the example that covers Parameters c. add Batch Processing d. improve the README
a. Add Idempotency
As part of this issue we should add something about the Idempotency utility. Based on the current sample I think the most logical place to add the utility would be in the put-item function as specifically modify the handler in such a way that the Idempotency utility is used to ensure that the same item is inserted only once.
Currently the code that inserts the item is part of the handler and looks like this:
const body = JSON.parse(event.body);
const { id, name } = body;
await docClient.send(
new PutCommand({
TableName: tableName,
Item: {
id,
name,
},
})
);I propose to extract the logic above to its own function and use the makeIdempotent function wrapper on it using all or some of the arguments.
b. Refactor Parameters
A couple of months ago we added the Parameters utility to the examples in conjunction with the GA of the utility. At the time I recall us struggling to find a way to use it and eventually we settled on creating a second REST API endpoint and using the utility to retrieve an arbitrary value which is then returned in the response.
After looking at the usage again, and also listening customer feedback, I think there's an opportunity to improve this area and use the utility for something more useful.
My first choice would have been to store the name of the Idempotency table (from the previous point) in a SSM parameter and then use the Parameters utility to retrieve it. This however is not possible at the moment given that we must instantiate the DynamoDBPersistenceLayer outside of the function handler and using await at the top-level requires ESM.
For context, and to maybe revisit this once we finally support ESM in the coming months, this is what I mean:
import { getParameter } from '@aws-lambda-powertools/parameters/ssm';
import { DynamoDBPersistenceLayer } from '@aws-lambda-powertools/idempotency/dynamodb';
const persistenceLayer = new DynamoDBPersistenceLayer({
tableName: await getParameter('/dev/idempotency-table-name');
});With that in mind, I think we could revisit this in a month or so and hopefully we'll be able to use the Parameters utility to retrieve the name of the Idempotency table using ES Modules.
c. Add Batch Processing
Given that we are inserting items in a DynamoDB Table, we could enable the Stream on the table and plug a fourth Lambda that consumes the stream and uses Batch Processing to do so.
The trigger for this function could be set to new items (aka inserts NOT updates / deletes). This function would provide an async record handler that would process each record in the batch and do something on it (TBD). Ideas:
- enrich the item with additional data
- do some validation and optionally reject the item (aka mark it as failed)
- others
d. Improve the README
At the moment the README is a bit bare and I think we could improve it by adding a couple of things:
- contextualize the example and explain what it does
- provide an overview of the architecture + diagram
- try to find a way to make the README more engaging by talking about a fictional use case that the example is trying to solve
@am29d what do you think? If we agree I'll move this out of the GA milestone.
AllyMurray
github-actions[bot]
Description of the improvement
Summary of the proposal The repository contains - at present - examples in CDK and SAM that showcase how to use the Powertools utilities available. The new idempotency utility should be part of these example.
How, where did you look for information In the CDK & SAM examples.
Missing or unclear documentation N/A
Improvement Include examples of the new utility.
Related existing documentation
https://github.com/awslabs/aws-lambda-powertools-typescript/tree/main/examples
Related issues, RFCs
447