One of the most common issues with the service has been addressed by an update to the serverless platform. When a function isn't being used for a long time, the virtual machine is usually shut down by Lambda. The idea is to create snapshots of a customer's functions and then start them without having to go through the usual process.
In today's keynote, the senior VP of Utility Computing noted that spiky workloads are pretty much what Lambda was built for, and that cold start times have always been one of the biggest complaints about the platform. The cold start times have improved with the Firecracker microVMs. The company says it will improve cold start times by using Firecracker.
This new feature can only be used for Java functions that use the Corretto runtime, and only for existing Lambda functions.
When you first run that function, it will do a standard initialization. A snapshot of the memory and disk state will be created after that. When the function is invoked again, the cache will be grabbed and the function started. After 14 days of being inactive, cached snapshots are removed.
More users will be able to bring their workload to a platform like Lambda as a result of this improvement. He said that the company saw this with the launch of Firecracker.
All Rights Reserved © 2024
