LAS VEGAS – Amazon Web Services (AWS) boosted its already industry leading Lambda serverless infrastructure platform with a new SnapStart quick start feature that targets laggy cold start issues.

Peter DeSantis, SVP of Utility Computing at AWS, explained during a keynote address at this week’s re:Invent 2022 show that SnapStart basically takes a snapshot of the memory and disk state of a running Lambda function. These snapshots are typically just a fraction of the full running function code and allows the platform to store a large number of these snapshots very efficiently.

SnapStart caches this snapshot in a micro virtual machine (microVM) to more quickly spin up a new Lambda function so that the new function does not need to start from the beginning of a cold start, which can often take several seconds to complete.

DeSantis said SnapStart “reduces cold starts by up to 90%, making cold starts virtually indistinguishable from cache hits.”

This is important as serverless platforms are typically used to quickly instantiate application functions that run for a short period of time. These are functions that may only last for seconds and need to be agile enough to spin up and down quickly.

SnapStart is built on AWS’ Firecracker virtualization service that it unveiled in late 2018. Firecracker is a virtual machine (VM) monitor that uses the Linux Kernel-based Virtual Machine (KVM) to create minimalist-designed microVMs and pack thousands of these microVMs onto a single machine to provide a smaller footprint for faster startup times and a reduced attack surface.

DeSantis added that VMs are typically random in how they use data to boot, but that this randomness is predictable.

“With SnapStart we profile how the virtual machine accesses every snapshot during its boot sequence and then we use this information to predictively load the chunks the next time it boots,” DeSantis explained. “This means that microVMs are never blocked, waiting for a snapshot to load. And this is what makes SnapStarts so fast.”

SnapShot maintains this uniqueness and security by requiring a user to verify that any unique content previously generated during the initialization is what gets generated after that initialization. This includes unique IDs, unique secrets, and entropy used to generate pseudo-randomness.

The SnapStart platform is being included for free within AWS’ Lambda platform.

The Lambda update should boost what is AWS’ already dominating position in the serverless platform space.

A recent report from Datadog noted AWS remains the benchmark for serverless usage. This includes the cloud giant’s Lambda and Fargate platforms. Fargate is a runtime layer for AWS’ EC2 that allows customers to run containers without managing servers and clusters.

Microsoft offers serverless and functions-as-a-service (FaaS) through its Azure Functions and Azure Container Instance (ACI) products, and Google provides similar services with its Google Cloud Functions and Cloud Run platforms.

Cisco’s most recent “2022 Global Hybrid Cloud Trends Report” found that 40% of enterprises surveyed that were using cloud-native technologies were using serverless. This was slightly higher than those that said they were using Kubernetes.

However, a survey earlier this year from the Cloud Native Computing Foundation (CNCF) found a slowdown last year in the adoption of serverless architectures. That survey of more than 19,000 software developers found the number of developers working on serverless architectures slipping from 27% of all developers in 2020 to 24% in 2021.

Photo Credit: AWS