Owen Taylor, Sr. Director, Gigaspaces

As Sr. Director, Worldwide Technical Communications with GigaSpaces Technologies Inc, Owen translates the new architectural concepts and technical capabilities of space-based solutions into accessible formats so that technologists can adapt them rapidly into their environments and gain their maximum benefit. Owens' areas of expertise include J2EE design patterns and performance tuning of J2EE applications.

Prior to GigaSpaces, Owen worked as Principal J2EE Product Specialist with Identify Software. Before that Owen acted as Senior Enterprise Architect with The Middleware Company where he specialized in B2B, EJB and J2EE training and consulting with a special emphasis on webMethods B2B server and, BEA WebLogic Servers. Owen has over the years delivered architecture consulting, mentoring and training to dozens of companies and advised them on how to best architect new applications ranging from e-commerce to stock-trading.

Many of his engagements involved developing application prototypes on-site. Prior to The Middleware-Company, Owen was Senior Consultant and Partner in The New Customware Company, where his duties mirrored almost exactly those he executed with the Middleware Company. Prior to CustomWare, Owen was Senior Consultant and Instructor in the Professional Services organization at Inprise (Borland) (an EJB/J2EE & CORBA vendor), where he provided consulting and mentoring to customers in not only building large applications with EJB/J2EE and CORBA, but also specifically on the instrumentation, monitoring and management of applications developed using these technologies.

Presentation: "No More Hops! - towards a linearly scalable application infrastructure"

Track: Solution Track: Architecting for Performance & Scalability

Time: Wednesday 16:00 - 17:00

Location: Concordia

Abstract:

Tier-based architectures are essentially database-based and this causes scalability issues both directly and indirectly.

Due to the heavy reliance on the database as the sole reliable storage mechanism, traditional architectures include mapping effort (ORM) as well as network hops to and from the DB which both cause increased latency. Additionally, traditional systems often utilize messaging as a stop-gap measure to defer processing that could be best performed immediately. As the demand on the number and size of requests supported by the application grows the pain caused by these two design decisions grows.

This talk focuses on an architecture utilizing Transparent Partitioning and Colocation. This allows the creation of a "processing unit", a single process that addresses reliability, messaging, and processing without involving network hops within the critical transaction flow of the application.

Learn how this pattern can be leveraged to increase throughput and reduce latency for Java, .NET, and even Excel applications.

Password protected Download slides