arrow_back Deep Learning for Natural Language Processing
Designing distributed components in a multi tenant architecture
Submitted by Ronak (@ronak-kothari) on Monday, 15 June 2015
The objective of this talk is to go over design of distributed search components in a Multi-tenant architecture spanning across geographies and deals with challenges around custom ranking, tenant specific configurations and dynamic ranking elements.
A global search infrastructure that serves millions of queries with SLA’s of the order of milliseconds is non-trivial to build. Throw in multiple tenants with diverse requirements and we have a huge complex system that needs to handle myriad configurations, ranking components, dynamic elements and what not. At stake of this complex system are stability, real time latencies, configurability, customization and scalability.
This talk will go over Bloomreach’s innovative search solution that addresses these seemingly impossible but utterly real requirements and the challenges involved in developing distributed search applications in a multi tenant scenario.
Some understanding of a search subsystem and how ranking works
Suchi and Ronak have been working on Search platform scaling for Bloomreach’s big data. Relevant experience includes distributed computing, data modelling, statistical machine learning, data ingestion, user data and personalization.
Ronak has an M.Tech, Computer Science from IIT, Bombay and BE from GCET. He worked at Adobe as a computer scientist for 7 years before joining Bloomreach.
Prior to Bloomreach, Suchi was a founding member of the Android framework team and has made extensive contributions to the Android framework.Suchi has an MS from University of Cincinnati and BE from BITS, Pilani. Her prior work experience includes working in companies like Google, Motorola and IBM Almaden.