Skip to content

Project Information Environment

Brian Hauer edited this page Nov 20, 2019 · 14 revisions

Simulating Production Environments

For this project, we aim for each framework to be configured according to its best practices for production deployments. Because the specifics vary by framework, we especially appreciate review from maintainers of each framework to confirm that the implementation(s) we have are indeed compliant with best practices.

The goal is to approximate and measure a sensible production-like deployment of each framework.

Environment Details

For language and framework specific details, view their README within their respective directory.

Hardware

Citrine (Self-hosted)

  • BENCHMARK_ENV=Citrine
  • Switched 10-gigabit ethernet

All 3 machines: tfb-server, tfb-database, tfb-client

  • 28 HT Cores Intel(R) Xeon(R) Gold 5120 CPU @ 2.20GHz
  • 32GB RAM
  • CentOS 7 kernel 3.10.0-957.21.2.el7.x86_64

Note: Hardware details used for the various rounds can be found on the results website.

Azure

  • BENCHMARK_ENV=Azure

All 3 machines: tfb-app-vm, tfb-data-vm, tfb-load-vm

  • Standard_D3_v2 (4 vcpus, 14 GB memory)
  • Default 30 GB of managed disk size (Standard HDD LRS)
  • TFB run and Docker images both utilize the temporary storage (under /mnt) of the Standard_D3_v2, which has 200 GB of SSD for the temporary storage
  • The network interface for each of the 3 VMs has the accelerated networking option enabled
  • The 3 VMs communicate via a virtual internal network

More information coming soon...

Load simulator

Clone this wiki locally