Lucene/Solr Revolution 2017 has ended
View analytic
Thursday, September 14 • 10:00am - 10:40am
SolrCloud Consistency and Recovery Internals
SolrCloud is a distributed search system, but how does it ensure that replicated data remains consistent? How does Solr avoid data loss when hardware inevitably fails? And how do temporary outages affect the availability of services?

In this talk, we will begin with a foundation of the relevant replication and leader election details before moving on to cover how Solr addresses failures and what recovery steps the cluster can automatically perform. We will examine the different classes of failure to better understand which ones are recoverable automatically, and which ones will require manual intervention by an operator. And in doing so, we will highlight which manual steps are safe and which ones carry more risk, as well as potential future improvements to the automated processes.

avatar for Mano Kovacs

Mano Kovacs

Software Engineer, Cloudera
Mano has been an application developer for more than 15 years. In the past years he was focusing on distributed, large scale services. He was working on an IoT platform before he joined the Search/Solr team at Cloudera in 2016.

Thursday September 14, 2017 10:00am - 10:40am
Banyan AB