AWS Lambda Unlocks Long-Running Serverless Tasks

AWS Lambda functions can now run for up to 90 minutes, a major increase that unlocks new use cases for long-running serverless tasks.

Amazon Web Services has significantly extended the maximum execution time for its popular serverless computing service, AWS Lambda. As reported by InfoQ, functions can now run for up to 90 minutes, a sixfold increase from the long-standing 15-minute limit. This substantial change redefines the boundaries of what is possible with serverless architecture, pushing it further into the territory of long-running computational tasks. The update specifically applies to functions deployed on what AWS refers to as Lambda Managed Instances, targeting asynchronous and event-driven workloads. It's a pivotal moment for the platform, which has historically been associated with short-lived, stateless operations. However, AWS has clarified that this new, longer timeout does not apply to all use cases; the execution limit for traditional synchronous requests, such as those triggered directly via an API Gateway, remains unchanged. This distinction underscores a strategic effort by AWS to expand Lambda's capabilities for a specific class of background processing jobs, rather than universally altering the nature of all function invocations.

The key to understanding this update lies in the type of workload it's designed for. The 90-minute runtime is not intended for the typical, immediate request-response patterns that power many web APIs. Instead, it targets asynchronous tasks that run in the background, such as processing a large file uploaded to an S3 bucket, executing a complex data transformation pipeline, or running a machine learning inference job. These are processes where a user isn't waiting for an immediate reply. Previously, developers needing more than 15 minutes for such a task had to architect complex workarounds. Common solutions included breaking the job into smaller pieces and orchestrating them with AWS Step Functions, or moving the workload entirely to a different service like AWS Fargate or a dedicated EC2 virtual server. This update simplifies that architectural choice dramatically. By allowing a single Lambda invocation to handle the entire job, developers can build and maintain simpler, more streamlined systems without the overhead of managing container clusters or virtual machines for these specific types of long-running processes.

This extension of Lambda's runtime is part of a broader industry trend where serverless platforms are evolving to handle a wider variety of computing patterns. When serverless first emerged, its primary value proposition was handling ephemeral, event-triggered tasks at scale without server management. This made it ideal for microservices, API backends, and simple data processing. However, its strict limitations, particularly around execution time and state management, kept it from being a viable option for more traditional, monolithic, or long-running batch workloads. Cloud providers are now actively working to dismantle these barriers. By increasing timeouts, offering better state management options, and improving performance, they are blurring the lines between serverless functions and more conventional compute services like containers and virtual machines. This move by AWS is a direct response to customer demand for simpler ways to run heavier workloads and a competitive play to keep Lambda at the forefront of serverless innovation, ensuring it remains a versatile tool rather than a niche solution.

For technology leaders and engineering teams, the practical implication of this change is immediate: it’s time to re-evaluate existing architectural patterns for long-running jobs. Workloads that were previously off-limits for Lambda, forcing the use of more complex and costly infrastructure, are now prime candidates for migration. This includes tasks like generating large reports, processing video files, or executing scientific computing simulations. Adopting Lambda for these use cases can lead to significant operational simplification and potential cost savings, as teams no longer need to provision, scale, and maintain dedicated servers for processes that run intermittently. Looking ahead, we can expect this trend to continue. It is likely that AWS will introduce further enhancements to support more complex applications on Lambda, potentially increasing other limits related to memory, storage, or concurrency. Teams should start experimenting with the new 90-minute limit for suitable background tasks, as it represents a fundamental shift in how serverless can be leveraged to build powerful and efficient cloud applications.

For developers, this update removes a major limitation of serverless architecture, making AWS Lambda a viable option for batch processing and ML inference jobs that were previously too long. It challenges the need for dedicated servers or container services for these specific workloads.

Companies can now consolidate more workloads onto a single serverless platform, potentially reducing operational overhead and infrastructure costs. This shift allows businesses to leverage serverless for more complex, long-running tasks without managing traditional servers or container orchestration, simplifying their tech stack.