Caching in the Clouds: Applying Ehcache and Terracotta to Cloud Deployments

Abstract:

Caching in the cloud is both a more important problem and is harder to solve than usual. CPU and I/O is lower, storage is slow and a database may be very slow. Fair resource allocation is not guaranteed, reliability and safety guarantees are opaque and there are proprietary rather than open APIs. Restrictions on multicast and even socket creation significantly curtail many approaches. This session will use source code and measurement tools to demonstrate the effects of caching in the EC2 and Google App Engine cloud environments.

In this talk, we will show how to use Ehcache, Hibernate, Quartz, Tomcat (or other containers) in conjunction with Terracotta to get to hundreds of nodes without all the separate stages of architecture and design such as messaging, partitioning, NoSQL, and more while avoiding cloud limitations. This recipe for scaling from 1 node to 1000 provides a seamless continuum within which your applications can operate and scale. This scalability continuum has implications on EC2 and EC2 has implications on the continuum. In this talk, we will also cover the ramifications of cloud on large-scale data management and we will apply the scalability continuum to our cloud-based applications to lessen our dependence on the database and, thus, take full advantage of the cloud's elasticity.

This talk is intended for software engineers that want to learn how to apply caching to cloud deployed applications.

In this talk, you will learn about:

1. Intro to Cloud
2. When it Rains: Cloud Problems
3. Google App Engine Caching restrictions and options
4. EC2 Caching restrictions and options
5. Ehcache + Terracotta as a scalability solution that alleviates your need for complex database infrastructure
6. EC2 and VMWare best practices for scaling with Terracotta's frameworks
7. Caching Topology Options: L1, L2, L1+L2, Grid, Sharding
8. Standard APIs: JCache and rest-*
9. The CAP Theorem and approaches to cache consistency: ACID, BASE, None
10. Terracotta's cloud-capable runtime and how to actually deploy to hundreds of nodes in hours
11. Example App and Demo in EC2: app using Spring, JPA, MySQL/Derby, Ehcache, Quartz, Terracotta
12. Conclusions
13. Q&A