Conference: 'Graphs at scale: lessons from building graph systems at LinkedIn and Microsoft'

Don't miss the conference about Graph Systems lead by Bogdan Arsintescu, Partner Director of Engineering at Microsoft.

Abstract

Graph Systems are essential for LinkedIn, with its over 1B member social network. LinkedIn has built bespoke scale-out graph systems throughout its existence to enrich its social network data and provide value to its members. 

We showcase the graph systems custom built at LinkedIn. We summarize the lessons we learned and the evolution of graphs over the last 15 years and three generations of graph systems. We discuss how the analysts, data scientists and developers at LinkedIn interact and benefit from these systems, how we provide the scale, availability and velocity of change required to build software to support the 1B LinkedIn members. The current graph is over 280B edges, with a throughput in excess of 2M QPS and “four nines” availability; the size and throughput have doubled in the last 18 months and the trends appear to accelerate. We outline how these solutions are leveraged to build Enterprise Graph solutions for the cloud.

Biography of the Speaker

Bogdan Arsintescu leads the Fabric Graph team in Microsoft, building scale-out graph systems for analytics and operational workloads. Prior to that he led the team that built and operates LIquid, a scale-out, low-latency, high-throughput and availability operational graph database system for LinkedIn. In the early 2010s Bogdan worked in the Google Knowledge Graph team; he led the Pregel-based high-performance compute graph processing system and is a co-author of PathQuery, the Google KG query language.

Join us next Thursday, 21st of March, at 12.00 h
Place: Sala d'Actes Manuel Martí Recober (B6) FIB

The event can be followed on our Youtube live stream