Not just u.
All our ci/cd are impacted
06.02.2025 08:56
👍 1
🔁 0
💬 0
📌 0
Not just u.
All our ci/cd are impacted
I think I can update my profile.
We committed to reduce this numbers, but it’s still going up slowly.
Still have a lot of things need to do .
Good news. 😭 it’s can save us thousands dollars per month.
aws.amazon.com/about-aws/wh...
Events driven on K8S or(Lambda on K8S) is more close to BEAM actor pattern.
Each container is only triggered by event (HTTP/SQS/Kafka, etc)
Similar with each gen_server get a message. For K8S, it’s not related to any language.
Erlang Community had a research to support partial cluster network or p2p style network. The big cluster problem is all the Beam nodes are connecting to each fully. Just keeping all hundreds nodes to ping each others to keep alive create a ton of loads on network and node itself.
DynamoDb can integrate with kinesis out of box. It’s a new feature in this year. IMHO, it’s better.The DynamoDb stream has a quota limitation. Per table only can have two steams.
Before it, we have to hook a lambda to publish it to a kinesis, so our data team and backend can consume it in both
For the light events, there is another pro is the schema compatibility. It’s kinda often the publisher will update the new event payload, adding new fields or add new enum items, etc. or remove fields. The light one have less fields and less chance to remove fields.
That’s nice
But when 🙂 they can support find the root cause within nested stack. Previously we have to scroll down many pages and jump to 5 depth nested stack to find the root cause.
Many developers gave us in the first page and throw the issue to infra or platform team to find it
SAM application for lambda?
What a cave age. Why we are still using YAML? AWS have a better one, CDK! Why can we just support it CDK at the first place.
Configure a lambda to consume a SNS topic with FIFO SQS + a DLQ those standard pattern in CFN template is a huge pain!