SolrCloud with large data sets and collections usually run into unevenly balanced clusters. This causes skewed data and replica distribution across SolrCloud nodes. Automatic Node Discovery in an existing cluster does not cause data to be re-distributed. Dynamically scaling up or scaling down collections based on index/config size or cluster size is non trivial and poses operational overhead.
Rebalance API offers a flexible way of redistributing data in SolrCloud while guaranteeing zero downtime. It offers multiple scaling strategies that aid in smarter/faster index manipulation and multiple allocation strategies that offer smarter collection placement in the cluster. It also offers a platform for users to write their own scaling strategy as they see fit. The api has been built in an open source friendly fashion.
Suruchi is currently working on BloomReach's Commerce Search Platform. She recently graduated from her master's program from Carnegie Mellon University focusing work on Big Data Machine Learning problems, Information Retrieval and Natural Language Processing. Her past experience includes... Read More →
Nitin has been working on solr based search platforms for over 6 years. His specializations include Distributed Systems, Performance Engineering, Search Platform.
Thursday October 13, 2016 10:50am - 11:30am EDT
Gardner
