Abstract:FaaS allows an application to be decomposed into functions that are executed on a FaaS platform. The FaaS platform is responsible for the resource provisioning of the functions. Recently, there is a growing trend towards the execution of compute-intensive FaaS functions that run for several seconds. However, due to the billing policies followed by commercial FaaS offerings, the execution of these functions can incur significantly higher costs. Moreover, due to the abstraction of underlying processor architectures on which the functions are executed, the performance optimization of these functions is challenging. As a result, most FaaS functions use pre-compiled libraries generic to x86-64 leading to performance degradation. In this paper, we examine the underlying processor architectures for Google Cloud Functions (GCF) and determine their prevalence across the 19 available GCF regions. We modify, adapt, and optimize three compute-intensive FaaS workloads written in Python using Numba, a JIT compiler based on LLVM, and present results wrt performance, memory consumption, and costs on GCF. Results from our experiments show that the optimization of FaaS functions can improve performance by 12.8x (geometric mean) and save costs by 73.4% on average for the three functions. Our results show that optimization of the FaaS functions for the specific architecture is very important. We achieved a maximum speedup of 1.79x by tuning the function especially for the instruction set of the underlying processor architecture.

Functional Augmented State Transfer (FAST) Architecture for Computationally Intensive Network Applications

Fast: A Simple Programming Abstraction For Complex State-Dependent Sdn Programming

FastScale: Fast Scaling out of Network Functions

Fastpath-based VPN Gateway over Network Processor

FAST TCP: motivation, architecture, algorithms, performance

A Survey on Architectures, Hardware Acceleration and Challenges for In-Network Computing

FastReact: In-Network Control and Caching for Industrial Control Networks using Programmable Data Planes

An efficient architecture for processing real-time traffic data streams using apache flink

In-Network Computing With Function as a Service at the Edge

Architecture-Specific Performance Optimization of Compute-Intensive FaaS Functions

Computation of an efficient pipelined fast Fourier transform architecture characterized with real-valued functions

FastFlow in FPGA Stacks of Data Centers

Swift: Fast, Reliable, Loosely Coupled Parallel Computation

Multitenancy for Fast and Programmable Networks in the Cloud.

Exploring the Approaches to Data Flow Computing

Improving the interoperability of a Function‐as‐a‐Service platform using an orchestration framework with a cloud‐agnostic approach

iFast: Host-Side Logging for Scientific Applications

Floe: A Continuous Dataflow Framework for Dynamic Cloud Applications

UniFaaS: Programming across Distributed Cyberinfrastructure with Federated Function Serving

SPATO: A Student Project Allocation Based Task Offloading in IoT-Fog Systems

Elastic high-performance computing platform for real-time data analysis