Videos
How Analytics Scale for Enterprise Workloads: Fast Data and Smart Caching
Description
What keeps analytics fast when hundreds of users run thousands of queries at the same time? It comes down to architecture: a high-performance data layer, smart caching, an intelligent computation engine, and cloud-native scaling working together.
In this session, we walk through how GoodData.AI handles enterprise-scale workloads, combining an Arrow-native data layer, multi-stage caching, and a computation engine that scales on Kubernetes.
The conversation covers:
- Built for concurrency: how the platform supports hundreds of concurrent users and thousands of parallel queries.
- The Arrow-native data layer: how Apache Arrow, an in-memory columnar format, enables zero-copy data sharing and fast analytical processing, keeping latency low under high concurrency.
- Caching across the pipeline: how results are cached at several stages, including the raw cache from the data source, the post-processing result cache, and the ML result cache.
- Configurable cache behavior: how durability, retention, and eviction strategies, cache warmup, and adaptive placement on disk or in memory keep the cache reliable, even under resource pressure.
- The computation engine: how it builds an execution plan for each request, runs workloads concurrently across data sources, reuses workspace-level caches across users, and uses prebuilt aggregates.
- Cloud-native scaling: how all services run on Kubernetes and scale horizontally, so performance stays consistent during traffic spikes.
Watch this session to see the architecture that keeps analytics fast and consistent as users and workloads grow, without slowing down at peak load.
Does GoodData look like the better fit?
Get a demo now and see for yourself. It’s commitment-free.

