Skip to content

pacslab/serverless-performance-modeling

master
Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 
 
 
web
 
 
 
 
 
 
 
 
 
 

PyPI PyPI - Status Travis (.com) Libraries.io dependency status for latest release GitHub

Performance Modeling of Serverless Computing Platforms

Package, experimentation results, and other artifacts for our serverless computing performance modeling paper. Using the presented performance model, the serverless computing platform provider and their users can optimize their workload and configurations to adapt to each workload being executed on them. The presented model uses analytical model to calculate steady-state system characteristics.

Benefits

  • Works with any service time distribution (general distribution).
  • Calculates steady-state characteristics.
  • Is tractable while having a high fidelity.

Artifacts

Here is a list of different artifacts for the proposed model:

Requirements

  • Python 3.6+
  • PIP

Installation

pip install pacsltk

Usage

Check out the package documentation.

Examples

You can use the package as simple as the short code snippet below:

from pacsltk import perfmodel

arrival_rate = 100
warm_service_time = 2
cold_service_time = 25
idle_time_before_kill = 10*60

print("arrival_rate:", arrival_rate)
print("warm_service_time:", warm_service_time)
print("cold_service_time:", cold_service_time)
print("idle_time_before_kill:", idle_time_before_kill)

props1, props2 = perfmodel.get_sls_warm_count_dist(arrival_rate, warm_service_time, cold_service_time, idle_time_before_kill)
perfmodel.print_props(props1)

which produces an output similar to the following:

arrival_rate: 100
warm_service_time: 2
cold_service_time: 25
idle_time_before_kill: 600

Properties:
------------------
avg_server_count: 251.043927
avg_running_count: 200.148828
avg_running_warm_count: 199.987058
avg_idle_count: 51.056869
cold_prob: 0.000065
avg_utilization: 0.796622
avg_resp_time: 2.001488
rejection_prob: 0.000000
rejection_rate: 0.000000
------------------

License

Unless otherwise specified:

MIT (c) 2020 Nima Mahmoudi & Hamzeh Khazaei

Citation

You can find the paper with details of the proposed model in PACS lab website. You can use the following bibtex entry:

@article{mahmoudi2020tccserverless,
  author={Mahmoudi, Nima and Khazaei, Hamzeh},
  journal={IEEE Transactions on Cloud Computing},
  title={{Performance Modeling of Serverless Computing Platforms}},
  year={2020},
  volume={},
  number={},
  pages={1-15},
  doi={10.1109/TCC.2020.3033373},
  url_paper={https://ieeexplore.ieee.org/document/9238484},
  url_pdf={https://pacs.eecs.yorku.ca/pubs/pdf/mahmoudi2020tccserverless.pdf}
}

About

Package, experimentation results, and other artifacts for the serverless computing performance modeling paper.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published