2023_12_wenRisePlanetServerless2023_Reading_Notes

Title: Rise of the Planet of Serverless Computing: A Systematic Review

zoterolink: Rise of the Planet of Serverless Computing A Systematic Review(2023 - Wen et al.).pdf

Reading Notes

Import zotero note:

%% begin notes %%
%% end notes %%

Annotations

Following content are annotations from origin paper:

%% begin annotations %%

Imported: 2023-12-27 11:35 pm

background adopted to develop a wide range of software applications, including machine/deep learning [252, 268], numerical computing [227], video processing [74, 125], Internet of Things [144, 277], big data analytics [119, 130], etc background AWS Lambda [16], Microsoft Azure Functions [22], and Google Cloud Functions [35]. Moreover, there are also some available open-source serverless platforms, e.g., OpenWhisk [48] and OpenFaaS [45]. background serverless evolution [243], characteristic analysis of serverless applications [114, 115], developers’ challenges [121, 257], background application modelling [272], programming framework of specific applications [84, 277], multi-cloud development [219], stateful serverless applications [82], application migration [214], serverless economic [63], serverless dataset [120], technical debt conceptualization [163], testing and debugging [164], etc. background resource management [137, 200, 275], cold start performance optimization [67, 126, 198], function communication [143, 230], general programming framework [124, 140] background Serverless computing provides “Backend as a Service” (BaaS) and “Function as a Service” (FaaS) target key characteristics target (1) Functionality and no operations (NoOps) target (2) Auto-scaling background Horizontal scaling is to launch (i.e., scale-in) new function instances or recycle (i.e., scale-out) running ones, while vertical scaling is to add (i.e., scale-up) or remove (i.e., scale-down) the amount of computation and other resources from running function instances target (3) Utilization-based billing: target (4) Separation of computation and storage target (5) Additional limitations background

background Resource Management; General Framework; Cold Start Performance; Function Communication; Function Execution; Stateful FaaS

background Generalizability of application conversion approaches

background generic conversion tools for any application

background Cold start performance problem of serverless applications

background Performance variability of serverless functions

background Fine-grained resource configuration on application development

background Heterogeneous accelerator support

background Monitoring tools

background Representativeness and completeness of benchmark dataset

%% end annotations %%

Summary

The Problem

Key Ideas & Key insights

Key Mechanisms

Implementation

Key Results

Key Conclusions

Strengths

Weaknesses

Can I do better?

Takeaways

Other comments

%% Import Date: 2023-12-27T23:35:31.959+00:00 %%