The homepage contains the dataset, scripts, and deployment code used in this paper.
-
Research papers: In the file "Collected research papers.xlsx", there is detailed information about 59 research papers that we collected, including
- CSRankings category
- conference
- publication year
- paper title
- the labeled information regarding RQ1
-
Raw data: We make raw data of cold-start/warm-start performance publicly available. Please refer to the file "Raw data of performance results.xlsx". We use the response latencies of each serverless function with 50 repetitions.
-
Serverless functions: We have also curated and made publicly available a dataset about serverless functions for future performance testing and optimization of serverless computing.
- In the file "Executed serverless functions.xlsx", there is detailed information about 65 serverless functions, including
- input payload
- executed serverless platform
- programming language
- timeout time
- memory allocation size
- original code link
- functionality type
- corresponding research papers
- In the file "Executed serverless functions.xlsx", there is detailed information about 65 serverless functions, including
- Invocation scripts: Invocation scripts used in our study are in the directory "Invocation scripts", including the invocation code of AWS Lambda and Google Cloud Functions.
- "invokeFunctionScript-run1.py" and "invokeFunctionScript-run2.py" for AWS Lambda
- "invokeGoogleFunction.py" for Google Cloud Functions
- In the directory "Deployment Packages for Serverless Functions", there are deployment packages corresponding to Func1 to Func65.
- the code for each serverless function is packaged in a separate folder
- the short input payload for some serverless functions is recorded in the file "Executed serverless functions.xlsx"
- the long input payload for some serverless functions is stored in the directory "input", including the payloads for func20, func29, func54, func56, func57, and func60