[Paper Review] A Review of Serverless Use Cases and their Characteristics

Simon Eismann, Joel Scheuner+6 more|arXiv (Cornell University)|Aug 25, 2020

Cloud Computing and Resource Management32 references47 citations

TL;DR

This technical report analyzes 89 serverless use cases across four data sources to characterize general, workload, application, and workflow traits, highlighting AWS dominance, production adoption, and typical use cases.

ABSTRACT

The serverless computing paradigm promises many desirable properties for cloud applications - low-cost, fine-grained deployment, and management-free operation. Consequently, the paradigm has underwent rapid growth: there currently exist tens of serverless platforms and all global cloud providers host serverless operations. To help tune existing platforms, guide the design of new serverless approaches, and overall contribute to understanding this paradigm, in this work we present a long-term, comprehensive effort to identify, collect, and characterize 89 serverless use cases. We survey use cases, sourced from white and grey literature, and from consultations with experts in areas such as scientific computing. We study each use case using 24 characteristics, including general aspects, but also workload, application, and requirements. When the use cases employ workflows, we further analyze their characteristics. Overall, we hope our study will be useful for both academia and industry, and encourage the community to further share and communicate their use cases. This article appears also as a SPEC Technical Report: https://research.spec.org/fileadmin/user_upload/documents/rg_cloud/endorsed_publications/SPEC_RG_2020_Serverless_Usecases.pdf The article may be submitted for peer-reviewed publication.

Motivation & Objective

• Identify and classify real-world serverless use cases from diverse sources.
• Characterize general, workload, application, requirement, and workflow attributes of serverless workloads.
• Assess production deployment, open-source availability, and domain distribution of serverless use cases.
• Provide insights to guide platform design, adoption decisions, and future research.

Proposed method

• Systematic collection of 89 serverless use cases from open-source projects, white literature, grey literature, and scientific computing sources.
• Dual-reviewer characterization of each use case against 24 predefined characteristics.
• Calculation of inter-reviewer agreement (Fleiss kappa) and consolidation of conflicting annotations.
• Public Zenodo repository hosting the full dataset.
• Classification into general, workload, application, requirement, and workflow categories.

Experimental results

Research questions

• RQ1 What are the common platforms, application domains, and production statuses of real-world serverless use cases?
• RQ2 What workload, application, and workflow characteristics typify serverless deployments in practice?
• RQ3 How do open-source availability and cross-domain applicability distribute among serverless use cases?
• RQ4 What motivates organizations to adopt serverless (cost, scalability, maintainability, performance) and how frequently are latency/locality requirements present?

Key findings

• AWS is the dominant deployment platform (80% of use cases).
• Web services is the most common application domain (33%), with 55% of use cases in production and 40% business-critical.
• API, stream/async processing, batch tasks, and operations/monitoring are the leading application types (28%, 27%, 23%, and 20%, respectively).
• 55% of analyzed use cases are deployed in production; about 36% have no clear production evidence, and unknowns are treated as not in production.
• 53% of use cases are open source; 47% are closed source.
• 62% of use cases rely on storage or databases; other common services include API gateway and pub-sub.
• 81% of workloads are bursty; most executions are on-demand rather than scheduled; many are lightweight HTTP-triggered.