Description
Burla is an open-source distributed computing platform for Python. It runs a function in parallel across many machines in your own cloud, and the site describes it as a way to prototype locally and then run the same code at scale. The core is a single call: remote_parallel_map(fn, inputs) calls your function on every input, each on a separate VM. You can set the CPU, GPU and memory for each function, for example an A100 GPU for an embedding step or 64 vCPUs for an indexing step, and chain functions together to build pipelines. Burla is aimed at machine learning pipelines, AI inference, batch processing and jobs such as OCR of large document sets. A dashboard lets you monitor every function call, with counts of completed and failed calls and the peak CPU and memory used. The project is hosted on GitHub, and the site has documentation and a blog.…
Highlights
- One function, remote_parallel_map, for parallel work
- Each call runs on a separate VM in your cloud
- Choose CPU, GPU and memory per function
- Monitor every function call in a dashboard
- Open source on GitHub
Ratings & Reviews
No ratings yet
Be the first to rate Burla.
People also visit
Burla FAQ
What is Burla?
An open-source distributed computing platform for Python. You wrap a parallel step in one function, and it runs across many machines in your cloud.
How do I use it?
Call remote_parallel_map with your function and a list of inputs. The function runs on every input in parallel, each on a separate VM.
Can I choose the hardware for a step?
Yes. The examples on the site set a GPU type, such as A100, or a number of CPUs, such as 64 vCPUs, for each function.
What kind of work is it for?
Machine learning pipelines, AI inference, batch processing and similar Python workloads. The site's examples include OCR of large sets of PDF documents.
What are the best Burla alternatives?
Popular alternatives to Burla include Backend.AI, Slurm-web and Clura. See all Burla alternatives






