Welcome to Yeedu
Yeedu is a multi-cloud platform that orchestrates the lifecycle of Apache Spark as a managed compute resource on AWS, Azure, and Google Cloud. A single Yeedu deployment lets you provision, run, and govern Spark workloads across any of these clouds, without rewriting your data pipelines or stitching together vendor-specific services. One control plane provisions the machines, runs the jobs on them, and keeps several Apache Spark runtime versions available side by side, on whichever of the three clouds you happen to be running that day.
Why teams choose Yeedu
Yeedu is built by data engineers who've spent years operating Apache Spark for large enterprises. The platform is designed to remove the biggest pain points of running Spark at scale.
Stability
Most managed Spark services run on Kubernetes, and most managed Kubernetes services (EKS, AKS, GKE) run on nested virtualization. Every instruction is translated by the guest hypervisor before reaching the host, which adds CPU overhead and unpredictable tail latency for analytics workloads.
Yeedu doesn't carry that overhead. It directly orchestrates virtual machines and bare-metal hosts using a scheduler designed from the ground up for cloud Spark workloads.
Cost
Many Spark vendors charge a multiple of the underlying compute cost. Yeedu is built to minimise that markup so the price you pay closely tracks the resources you actually consume.
Multi-cloud and portability
You can move workloads between AWS, Azure, and GCP without rewriting your foundational code. This lets you take advantage of cloud-specific discounts today so you aren't locked in tomorrow.
Resource optimization
- Yeedu ships multiple schedulers, including one tuned for I/O-bound jobs, so you can match the scheduling strategy to the workload instead of running everything through a single generic scheduler.
- Running compute close to your data, possibly across more than one cloud, reduces cross-region and cross-cloud egress costs.
Multi-version Spark support
Yeedu ships several versions of the Apache Spark runtime side by side, so you can pin production jobs to a known-good version and migrate at your own pace. Each runtime version also declares which capabilities it supports, Notebook, CUDA, Turbo and Thrift, and you pick one per cluster.
Where to next
- New to Yeedu? Start with Get started for the end-to-end workflow.
- Want the high-level picture first? Read Architecture and Concepts.
- Ready to install? Jump to Set up Yeedu and pick your cloud.