Distributed Persistent Memory for Spark and TBD

Agenda:

6:00: Food/drinks arrive

6:20: Talk #1: Distributed Persistent Memory for Spark

7:00: Questions

7:15: Talk #2: TBD

8:00: Questions

8:15: chill + relax = chillax

Description

Talk #1: Distributed Persistent Memory for Spark

Abstract:

Recent commodity server advances in multi-core, multi-channel IO, and Flash storage have enable us to build a new, native type of indexed Key Value storage and analytics off load engine that is vastly more performant and lower latency than existing Log Structure Merge approaches. The result is a distributed memory-speed persistent storage and analytics off load engine that is ideal for use cases such as real-time streaming ETL, analytics, and messaging.

This technology is additionally been adapted/integrated with Apache Spark via a connector and we will discuss/demonstrate how this can supplement Spark users who would benefit from availability of Persistent Sharable Dataframes for concurrent user access and for streamlining overall multi-job workflow execution. Certain IO intensive and iterative operations and queries can be offloaded/greatly accelerated by this approach.

Bio:

Bernie is Chief Business Development Officer for Levyx (Irvine, CA) and passionate about their mission to deliver innovative software to foster Real-time Persistent Computing for Big Data. Previously, Bernie was at 2 Silicon Valley startups in the areas of Software-defined storage and Flash-Object storage. Bernie was also a Founder/EVP for FalconStor - a Software pioneer in storage virtualization and heterogeneous data replication Prior to that, he was SVP at Trend Micro where he pioneered building virus protection onto Internet email relays. He has BS/MS Eng. degrees from UC Berkeley and an MBA from UCLA.

Talk #2: TBD