In recent years, serverless computing or Function-as-a-Service (FaaS) solutions are growing in popularity and finding their way into commercial clouds (AWS Lambda, Google Cloud Functions, Azure Functions, etc.) and open source projects (OpenWhisk, OpenFaaS, etc.). While serverless platforms were originally intended for event-driven, stateless applications like image processing, a recent trend is the use of serverless computing for more complex, stateful, parallel, data-intensive and/or compute-intensive applications, like machine learning, scientific computing workflows, and data analytics. These large stateful applications have high parallelism and require scalable computing capability. Additionally, FaaS platforms are pay-per-use, thus eliminating potential waste on monetary cost. The elasticity, auto-scaling, pay-per-use properties provided by FaaS platforms makes them potentially well-suited for such kind of stateful applications.
Supporting large stateful applications on FaaS, however, introduces new challenges. First, while FaaS platforms promise to offer superior elasticity and auto-scaling properties, function invocations introduce non-trivial overhead. Unlike typical serverful computing frameworks where the scheduler directly communicates with workers using low-latency RPC requests, serverless functions can only be dispatched using high-cost HTTP. Second, serverless functions come with inherent constraints, including bandwidth-limited, outbound-only network connectivity, lack of data availability support, and limited computing and memory resources. Therefore, FaaS users and developers need to seek ad-hoc, and often inefficient solutions to walk-around these inherent limitations in order to support large stateful computing on FaaS platforms.
Topics of interest to this Research Topic include but are not limited to the following:
- New serverless computing applications
- New FaaS and serverless capabilities that enable efficient stateful computing
- New systems techniques (e.g., storage, network, etc.) to enable better FaaS service
- Big data management and processing on serverless and FaaS
- Serverless backend as a service (e.g., serverless databases, serverless storage)
- Energy and carbon efficiency of serverless
- Security, privacy, and trust management of serverless and FaaS
- Applications of machine learning and edge computing in serverless
Keywords:
Serverless, FaaS, stateful computing, parallel computing, distributed systems
Important Note:
All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.
In recent years, serverless computing or Function-as-a-Service (FaaS) solutions are growing in popularity and finding their way into commercial clouds (AWS Lambda, Google Cloud Functions, Azure Functions, etc.) and open source projects (OpenWhisk, OpenFaaS, etc.). While serverless platforms were originally intended for event-driven, stateless applications like image processing, a recent trend is the use of serverless computing for more complex, stateful, parallel, data-intensive and/or compute-intensive applications, like machine learning, scientific computing workflows, and data analytics. These large stateful applications have high parallelism and require scalable computing capability. Additionally, FaaS platforms are pay-per-use, thus eliminating potential waste on monetary cost. The elasticity, auto-scaling, pay-per-use properties provided by FaaS platforms makes them potentially well-suited for such kind of stateful applications.
Supporting large stateful applications on FaaS, however, introduces new challenges. First, while FaaS platforms promise to offer superior elasticity and auto-scaling properties, function invocations introduce non-trivial overhead. Unlike typical serverful computing frameworks where the scheduler directly communicates with workers using low-latency RPC requests, serverless functions can only be dispatched using high-cost HTTP. Second, serverless functions come with inherent constraints, including bandwidth-limited, outbound-only network connectivity, lack of data availability support, and limited computing and memory resources. Therefore, FaaS users and developers need to seek ad-hoc, and often inefficient solutions to walk-around these inherent limitations in order to support large stateful computing on FaaS platforms.
Topics of interest to this Research Topic include but are not limited to the following:
- New serverless computing applications
- New FaaS and serverless capabilities that enable efficient stateful computing
- New systems techniques (e.g., storage, network, etc.) to enable better FaaS service
- Big data management and processing on serverless and FaaS
- Serverless backend as a service (e.g., serverless databases, serverless storage)
- Energy and carbon efficiency of serverless
- Security, privacy, and trust management of serverless and FaaS
- Applications of machine learning and edge computing in serverless
Keywords:
Serverless, FaaS, stateful computing, parallel computing, distributed systems
Important Note:
All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.