An Empirical Case Study on the Temporary File Smell in Dockerfiles
Authors:
Open Access
- 15 March 2019
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 7, 63650-63659
Abstract
Docker is widely used in data centers to host services. The docker image adopts hierarchical storage architecture, which means that the docker image is composed of a set of filesystem layers. In the image building process, only the top layer is read-write while the bottom layers are all read-only. However, temporary files are often used in the image building process. Nevertheless, if a temporary file is imported and removed in different layers by a careless developer, it will lead a file redundancy. We termed this problem “temporary file smell”. This smell leads to larger-size images, which seriously restricts the efficiency of image distribution and thus affects the scaling ability of services in facing of sudden high load. To address this problem, we make an empirical case study to the real-world Dockerfiles on DockerHub. Based on the case study, we summarize four different smell patterns and propose a state-depend static analysis method to detect this kind of smells. We also provide three feasible fixing methods as selective options to eliminate the temporary file smell.Keywords
Funding Information
- National Key Research and Development Program of China (2017YFB1400804, 2017YFC0804407)
- National Natural Science Foundation of China (61602454, 61872344)
- Natural Science Foundation of Beijing Municipality (4182070)
- Youth Innovation Promotion Association of the Chinese Academy of Sciences (2018144)
This publication has 11 references indexed in Scilit:
- RUDSEA: recommending updates of Dockerfiles via software environment analysis2018
- An Insight Into the Impact of Dockerfile Evolutionary Trajectories on Quality and Latency2018
- Structured information on state and evolution of dockerfiles on github2018
- Docker-Based Automatic Deployment for Nuclear Fusion Experimental Data Archive ClusterIEEE Transactions on Plasma Science, 2018
- Cider: a Rapid Docker Container Deployment System through Sharing Network Storage2017
- An Empirical Analysis of the Docker Container Ecosystem on GitHub2017
- Clustering-based acceleration for virtual machine image deduplication in the cloud environmentJournal of Systems and Software, 2016
- Cobra: a light-weight tool for static and dynamic program analysisInnovations in Systems and Software Engineering, 2016
- Comprehensive Multiplatform Dynamic Program Analysis for Java and AndroidIEEE Software, 2015
- An introduction to Docker for reproducible researchACM SIGOPS Operating Systems Review, 2015