Machine Learning in the Cloud with Mahout and Whirr
summary
Shows how to launch mahout cluster on public clouds by having Whirr fetch and install the Mahout tarball, set env vars and mount EBS with Apache archives so MapReduce jobs run immediately. Demonstrates support for multiple providers via jclouds, a working example cluster for testing Mahout development, and a file-based instance map for SSH. Notes hard limits: no hot node addition yet, observed failures past ~20 nodes on some providers, and planned VirtualBox deployment to speed cycles.
AI-written summary · about
chapters (7)
- 0:00Start
- 0:32Agenda
- 5:22Deploy Services in the Cloud
- 8:29How Do You Start a Cluster
- 12:40Final Tips and Tricks
- 14:01Upcoming Features
- 15:08References
related talks
20mHow the cloud changed the way I do analytics and research
16mAndrew Musselman, Committer & PMC Member, Apache Mahout at MLconf Seattle 2017
31mRunning Machine Learning Analytics On Traces
01h 21mHPC in the cloud
34mOnline learning, Vowpal Wabbit and Hadoop #bbuzz
40mCloud-based Machine Learning Offerings - Beginner's Perspective
40mStreet fighting techniques for multi tenant machine learning...
44mApache Spark and ML workflows