LESS COMPLEXITY, MORE RESILIENCE, CLOUD-NATIVE

No Hadoop

Lentiq offers many of Hadoop's features while also being cloud native, more resilient and easier to use.

Get in touch

A better platform was needed

We love Hadoop, we really do. Our team here at Lentiq has been using it for many years for building big data lakes of all sorts. However,   the same years of experience also showed us what needs to be improved:

  • Less complexity
  • More cloud nativeness: auto-scaling, object storage etc.
  • A simpler security model
  • A better data catalog
  • A better workflow mechanism

Requirements have changed

Customers' needs have changed since Hadoop was invented in 2005. It was pretty much the dark ages back then when we used to call it "big data". Lentiq was built with these new requirements in mind:

  • Native model lifetime management - feature engineering, periodic model training, model serving require dedicated instrumentation
  • Better support for deep learning tools - Tensorflow, PyTorch bypass the map-reduce model entirely and leverage GPUs for performance
  • Better self-service support - Easier to use by data scientists, a better data catalogue, a simpler security mechanism, better sharing mechanisms

How do we do it?

Lentiq built on Kubernetes, on-demand cloud resources, object storage and Docker rather than Hadoop to leverage the cloud better and to provide a better user experience. However Kubernetes alone is not enough. Lentiq developed several unique technologies: 

  • LambdaBook Workflows - Easily convert notebooks into executable Docker containers and run them as part of workflows.
  • Cross-Cloud Data Store and Code Store  - Cross-cloud data, metadata and code discovery and sharing mechanisms.
  • Cross-cloud portability layer - Code that runs in one cloud will run on another without modifications.

Do you want to try Lentiq with your team?

Get in touch