Thanks for the cool insights! Another interesting take would be to compare and contrast distributed maps to Batch array jobs. I’d be super curious! 😊 (you can consider that serverless also when you use ECSFargate to run the containers)
Having worked on few services buses earlier like Oracle Service Bus and WebSphere ESB, when Event bridge service came out initially with the event bus, I could easily relate to the way we used to do traditional service bus, routing, transformation and all that. It has now come along way in adding so much capability. Love this service and really nice to see the internals of how this all stacks up. Really great discussion.
One missing piece is to have ultra low latency to communicate between lambdas… so a set of lambdas could work together as “almost” a single super computer. Like low latency in HPC
I have the NAT and everything setup for lambda, it can access the external database but cannot process httpclient or acquiretoken , it ends the process without error> private static readonly HttpClient client = new HttpClient(); HttpResponseMessage response = await client.GetAsync("www.google.com"); or var result = await _confidentialClientApplication .AcquireTokenForClient(OAuth2.scopes) .ExecuteAsync(cancellationToken);
A couple of questions about Step Functions - 1. We've got logging hooked into lambda output, so have been able to set up extensive reports on errors, etc through Cloudwatch reports. Converting lambdaa to Step Functions has been very difficult, due to the fact that the execution history is disconnected from Cloudwatch. We were able to send the execution history, but it's very difficult to control the output so that we can structure the reporting - collecting error stats with patterns and other processing stats. Is there any plans to work on improving the Step function logging to Cloudwatch so we can add observability to the results of calls to all the integrated AWS services? 2. Unit / Intergration / UAT testing frameworks are badly needed. AWS IATK is a nice start, but we've got all our objects / lambdas / code written in Node. What is the recommended way to do automated testing with step functions?
How would one handle the responses returned from a target API in an API Destination request? For example, a GET request returns data by definition. Otherwise, why would a GET request be of any use as an API Destination verb? Thanks!
Thank you very much. Sounds like a very useful technology to speed up development saving the time it takes to redeploy to the "real" AWS Cloud. Very useful for unit tests as well instead of mocking AWS services, using this "emulated" cloud locally will be more effective.
Thanks .. I have a query about runtime environment & lambda function. I understand Runtime Environment and Lambda Function has their own set of modules/libraries. Is there a way to know what packages/libraries etc used in runtime environment? When we build/deploy our lambda code, it has its own specific version of modules/libraries (like botocore, powertools etc). I assume runtime env may get upgraded by AWS from time to time But lambda function still has old version of libraries (unless we rebuild/deploy our lambda). Will that cause any error as module versions will be different or not compatible? What is the best way to avoid it?
I am needing to replicate a salesforce object, this would be perfect for my use case if salesforce also sent all of the columns in the payload that didn't have a change as 'null' is there anyway to set that in salesforce?
This is amazing. As someone who writes yaml & serverless files and bangs their head every time a new change breaks the code pipeline, this could be ground breaking if it works as demonstrated.
This is awesome thank you. I did want to clarify one point in the 17-18 minute mark. I believe app flow has improved their connector to salesforce and you can now stream change events from salesforce through app flow using an object event channel. I’ll still be using event bridge for the added features but if you need data from sf to dump to s3, redshift, or Postgres app flow can be set up in a minute. If only they had an msk destination.
I'm curious about the ids for customer, product, warehouse, etc. How do we ensure there aren't id collisions across entities? Are these ids UUIDs and therefore all entities can take their ids from the same "stream" of generated ids, and we don't worry about separate sequences of ids for each entity?
This is incredibly insightful! The flexibility unleashed by this integration is unmatched. With EventBridge's native integration with numerous AWS services, organizations gain the flexibility to seamlessly integrate with AWS. One of the greatest advantages is the elimination of managing glue code, previously required for this using SFDC pub-sub APIs and maintaining custom connectors. This native AWS and SAAS integration automatically handles these tasks, simplifying the process significantly.