Track: Sponsored Solutions Track I

Power of Graph Algorithms to Understand Your Data

Neo4j just released an open source library of graph algorithms that run directly in their open source graph database. These algorithms, callable from Cypher, enable clustering & community detection, centrality measurement, and more. Applying these algorithms, developers can create better content recommendations, improve social influencer analysis, and improve fraud detection.

This represents a new architectural pattern for graph analysis -- running real-time and directly on top of the OLTP graph data in memory vs exporting to an analytics solution like Spark+GraphX, as previously recommended.

In this session, we'll explore the performance characteristics of this approach and dive into example graphs and show how graph algorithms can be applied to them for fun and profit. Expect to code and some pretty graph visualizations!

The Serverless Application Lifecycle with Azure

Serverless frees developers from managing infrastructure, but it can be challenging to debug, deploy, and monitor applications. In this session, we'll demonstrate best practices for unit testing, integration testing and deployment. Learn how to create a robust CI/CD pipeline and deploy safely to production. Instrument your code so that you have rich telemetry and alerting. We'll show how to implement these best practices using various serverless technologies in Azure.

Event Driven Architecture for Real-Time Analytics

As the world is becoming more instrumented and connected there is a shift to event driven architectures that analyze and act on events as they happen to give more intelligent applications and personalized experiences. We’ll discuss the shift from a traditional big data approach to an event driven architecture for real time analytics and intelligent applications. The technologies and components that make up an event driven architecture and the changing demands of moving to real time. We’ll also explore use cases that show how enterprises are using these technologies at scale to solve real-world problems and exploit new opportunities.

I Have a NoSQL Toaster

My toaster stores data without SQL and without tables. But making a choice based on what something doesn’t have isn’t terribly useful. “NoSQL” is an increasingly inaccurate catch-all term that covers a lot of different types of data storage. Let’s make more sense of this new breed of database management systems and go beyond the buzzword. In this session, the four main data models that make up the NoSQL movement will be covered: key-value, document, columnar and graph. How they differ and when you might want to use each one will be discussed.

This session will be looking at the whole ecosystem, with a more detailed focus on Couchbase, Cassandra, Riak KV, and Neo4j.

Fix Spark Failures and Bottlenecks Faster & Easier

This talk presents the results of analyzing many Spark jobs on many multi-tenant production clusters. Kirk discusses common issues seen, the symptoms of those issues, and how developers can address them.

At Pepperdata, we have gathered trillions of performance data points on production clusters running Spark, covering a variety of industries, applications, and workload types. We will present key performance insights — best and worst practices, gotchas, and tuning recommendations — based on analyzing the behavior and performance of millions of Spark applications. In addition, we will describe how we are turning these learnings into heuristics used in the open source Dr. Elephant project.

YugaByte: Cloud-Native DB Converging SQL & NoSQL

Today’s public and private clouds are built to run cloud-native applications architected for shared-nothing, scale-out, commodity infrastructure that can fail often. They give enterprises easy access to a number of geo-distributed datacenters for increased availability and fault-tolerance but also require application developers to think about infrastructure agility, data locality and cross-datacenter replication. Mission-critical applications with their strict need for data correctness and availability have thus become significantly challenging to run on modern cloud infrastructure.

Current generation of operational databases were never built to solve the above challenges. Strongly consistent scale-out SQL databases, that started out as sharding and clustering layers on top of monolithic SQL databases (such as MySQL and Postgres), provide the much needed data correctness guarantees but are only fit for private datacenters interconnected with highly reliable custom networks. Eventually consistent NoSQL databases (such as Cassandra and MongoDB) provide higher availability but do not provide strong guarantees around data consistency or zero data loss during failures.

In this session, we will review the architecture and design of YugaByte DB, a new open source, cloud-native database purpose-built for mission-critical applications. YugaByte DB provides a strongly consistent core similar to SQL databases while also enabling multi-datacenter high availability similar to NoSQL databases. Stateful applications can now easily scale up and scale down in both public and private cloud platforms without incurring operational complexity or losing data during failures.