Building Resilient Distributed Java Systems: From Failure to Recovery

In modern Java applications, distributed systems are everywhere, and so are failure modes. But how do you know when your cluster is fragile, or if it’s on the brink of breaking?

This talk dives into practical observability and resiliency techniques for distributed Java environments. We’ll highlight key patterns, failure signals, and metrics that matter, backed by a live demo using Hazelcast, Chaos-mesh, Prometheus, and Grafana.

You’ll learn:

Core Patterns – Leader election, partitioning, replication
Metrics That Matter – Backup count, member count, JVM health, Golden Signals
Failure-Aware Design – Resilience patterns, chaos testing principles
Live Demo – Deploy a working cluster, simulate node failure, and explore metrics to observe how data integrity holds as the system nears its fault tolerance threshold

Ideal for Java developers, architects, and SREs, this session blends theory, tools, and real-world failure scenarios to help you build distributed systems that stay online—even when things go wrong.

About the venue
Free Times Cafe has bistro-style seating and a full food and drink menu. Please consider helping to support the venue by planning to have supper during the talk.

Speaker Bio
Joe Sherwin is a Principal Solution Architect at Hazelcast with 22 years of experience in the design, development, and implementation of application systems within multi-tier distributed computing environments. Working with clients such as Vanguard, Fannie Mae, Federal Reserve Bank, Citi Group, Bear Stearns, Fixed Income Clearing Corporation, Comcast Corp, Webster Bank, Gartner Group, The Hartford Life Company, IBM Global Services, Mass Mutual, Lincoln National Financial Corporation, Bank of America, and Barnes & Noble Online Group, Mr. Sherwin has been instrumental in the development of large-scale mission-critical E-commerce, insurance, and financial systems. He has experience architecting & implementing solution using CORBA, RMI, Java EE compliant distributed Object architectures, in-memory high transaction/low latency solutions using Hazelcast IMDG®, GemFire, Ehcache & Oracle Coherence, and solutions deployable on IaaS or PaaS platforms like Cloud Foundry, Amazon Web Services, Rackspace or Heroku.