September 30th, 2019 by Michael Rink
Cisco Releases Two CDIP Validated Designs Using Cloudera
Last Friday Cisco released two, nearly identical, new Cisco Validated Designs implementing Cisco Data intelligence Platform (CDIP). One of the designs runs Cloudera Enterprise Data Hub 6.2, and the other runs Hortonworks (recently purchased by Cloudera) Data Platform 3.1. Getting customers to be interested in the new systems is probably going to be an uphill battle for Cisco because Cloudera is moving to merge Hortonworks and their own software together. Cloudera has already released the first results of this merge, as we covered earlier this week.
Cisco Data intelligence Platform (CDIP) was introduced a few months ago (last June) and brings together big data and storage tiers in a single entity. The design objectives for CDIP architectures like the two just released are:
- Extremely fast data ingest to the data lake
- Computational elements allowing for different types of AI frameworks and compute types (GPU, CPU, FPGA) to work on this data for further analytics
- A storage tier which can gradually retire data which has been worked on to a storage dense system with a lower $/TB providing a better TCO
- Seamlessly scale the architecture to thousands of nodes with a single pane of glass management using Cisco Application Centric Infrastructure (ACI)
There are very few differences between the two Cisco Validated Designs (CVDs). Both designs share the same four basic elements: a data lake with Hortonworks Data Platform 3.1 or Cloudera Enterprise Data Hub 6.2; a Kubernetes / AI farm with Cloudera Data Science Workbench; a Hadoop Tiered storage with S3260; and a distributed AI/ML with Apache Submarine. The Performance version of both designs is identical.