Peeking Behind the Curtains of Serverless Platforms

By Liang Wang, Mengyuan Li, Yinqian Zhang, Thomas Ristenpart, Michael M. Swift. USENIX Annual Technical Conference, 2018

667 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/bbb606d78d379262c85c4615cb5d9c191cd2e3bf

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Serverless Computing: One Step Forward, Two Steps Back (2018, background): “With 20 Lambda functions, average network bandwidth was 28.7Mbps—2.5 orders of magnitude slower than a single SSD [26] 4 .”
SPINN: synergistic progressive inference of neural networks over device and cloud (2020, methodology): “Nevertheless, in a typical deployment, we would envision a caching proxy serving the models with RDMA to the CPU or GPU of the end server, in a virtualised or serverless environment so as to tackle the cold-start problem [57, 77].”
Faasm: Lightweight Isolation for Efficient Stateful Serverless Computing (2020, background): “Container initialisation times have been reduced to mitigate the cold-start problem, which can contribute several seconds of latency with standard containers [36, 72, 83].”
The rise of serverless computing (2019, background): “Evaluating the performance of different serverless platform to identify the tradeoffs has been a recent topic of investigation [17, 26, 27], and recent benchmarks have been developed to compare the serverless offering by the different cloud providers .”
SOCK: Rapid Task Provisioning with Serverless-Optimized Containers (2018, background): “Wang et al. [50] reverse engineer many design decisions made by major serverless platforms.”
Catalyzer: Sub-millisecond Startup for Serverless Computing with Initialization-less Booting (2020, background): “The ephemeral execution sandboxes are typically containers [1], virtual machines [20, 44] or recently proposed lightweight virtualization designs [6, 8, 19, 35, 37, 41, 45].”
Cloud Programming Simplified: A Berkeley View on Serverless Computing (2019, methodology): “ing relies on strong performance and security isolation to make multi-tenant hardware sharing possible. VM-like isolation is the current standard for multi-tenant hardware sharing for cloud functions [12], but because VM provisioning can take many seconds serverless computing providers use elaborate”
FaasCache: keeping serverless computing alive with greedy-dual caching (2021, methodology): “Using virtualization techniques, a single physical server can execute many functions concurrently and safely [55].”
Serverless in the Wild: Characterizing and Optimizing the Serverless Workload at a Large Cloud Provider (2020, background): “The main implication is that the function execution times are at the same order of magnitude as the cold start times reported for major providers [44].”