Cloud Traces

Traces from existing parallel and distributed computing systems are a useful resource for researchers to replicate real-life workloads for their experiments. However, since cloud computing is a relatively new area, few such traces are currently available. Chameleon makes available trace data from its operational OpenStack cloud infrastructure — including bare-metal and virtual machine workloads spanning over a decade — to support the broader research and education community.

Citing This Data

If you use Chameleon cloud traces in your research or coursework, please cite the dataset page from which you downloaded the data. Proper attribution helps sustain continued access to this resource and supports reporting to our funders. We also welcome you to reach out and let us know how you're using the data — you can find our contact information here.

Recommended Citation (Bibtex):

@misc{chameleon_cloud_traces,
  title        = {Chameleon Cloud Traces},
  author       = {{Chameleon Cloud Project}},
  year         = {2026},
  howpublished = {\url{https://chameleoncloud.org/data/cloud-traces/}},
  note         = {Updated: March 2026}
}

Using Chameleon Cloud Traces

We have developed a trace data structure based on data from OpenStack Nova/Blazar/Ironic services, as well as software to extract the appropriate data.

The following are the latest cloud trace files from Chameleon. Downloads are tracked for reporting purposes.

These datasets are updated quarterly (using open-source tools published on GitHub). Next data update: July 1, 2026.

Note on machine events: Current trace files do not include historical machine events for active nodes. Researchers needing this information — for example, details on node configurations or lifecycle events — may find additional data in the earlier ScienceClouds traces, which also cover nodes that have since been retired. We plan to expand machine event coverage in future releases.

Getting Started

For a quick start, we provide a Jupyter Notebook example that you can upload to the Chameleon Jupyter server. For more information about the Chameleon Jupyter interface, see the Chameleon user documentation.