Nati Shalom's Blog

Intersting analysis about GigaSpaces and Space Based Architecture

As a CTO I often find myself about positioning and technology direction for GigaSpaces. I must admit that with the rapid changes in middleware and technology it is a growing challenge, which requires continuous adjustments and re-alignment. Unlike many of the alternative solutions, we chose to take a relatively controversial direction on several fronts. For example, the "End of N-Tier" approach in which we argue that the separation to physical tiers is artificial and leads to many inefficiencies, complexity and lack of scalability; or the "putting your database where it belongs" argument in which we assert that the role of the database should be diminished to a background persistemt layer, and instead you should use an in-memory data-grid as the system-of-record. At the same time we worked hard to make these different approaches as seamless as possible to developers, especially those with JEE or Spring experience.

As someone who has been advocating these concepts for some time, I find it encouraging to see that the message is slowly but surely getting across. The most recent indication is an analysis To scale out or not using Gigaspaces  written by Anand Ganesan. I particularly liked the following paragraph that I thought provide a good description of our XAP value proposition.

"The product XAP is an infrastructure software platform providing full scale out capability to applications. XAP is compatible with JEE. With JEE support, Gigaspaces wants a share in the Java application server market. For example, XAP can execute a Java application instead of executing it in a traditional application server such as Weblogic. The motivation to this rationale (from a technical viewpoint) is that, performance and scalability bottlenecks are not always limited to database, it may be due to communication mechanisms such as, for example, messaging between applications or in the application server tier itself. Thus end to end latency issues can be addressed, by using XAP. Some vendors (Appistry, IBM Websphere XD, GridGain, and Data Synapse) use a grid computing approach to overcome scalability issues for high performance computing applications. However XAP uses a Space Based Architecture (SBA) approach borrowing from the Java Spaces Specification. The challenge to adopting SBA architectural style is the change in the mindset from using the traditional tiered based architectural patterns. However, arguably, the benefits of using SBA are in availing linear scalability, high availability, and simplified infrastructure among other things, which outweigh the benefits of traditional tiered based architectural styles."

Enjoy!

GigaSpaces Annual User Forum Event - London - September 22, 2008

Our second customer event will take place on the 22nd of this month.  Judging by the agenda and the list of presenters and attendees, it looks like this event is going to continue the successful tradition of last year's event, where various customers presented their stories and shared experiences. If you are interested, you can get a feel for it by looking at stories from last year's event.

This year we are planning to have a wide variety of customer case studies from verticals such as Energy, Finance, and Telco. RWE will begin by speaking about The Use of GigaSpaces within the Energy Trading Industry, and how their work with GigaSpaces leverages many of our new product features, such as .Net and C++ interoperability, dynamic language support, and more.

As a GigaSpaces' Systems Integrator, Avenue A Razorfish will speak about their experience in building high performance and reliable SOA framework for order management systems for global mobile telecommunication companies. These systems have very strict reliability constraints and built-in support for on-demand scaling to meet peak loads.

I'm also very happy to announce a case study by Nortel, one of our first customers, using GigaSpaces for several years in production for their contact center application - essentially one of the classic Event-Driven applications. Another session worth noting during the day is Mathew Fowler's session, which will demonstrate an integration between JeeWiz and GigaSpaces. In this integration, Matthew will demonstrate how to build a GigaSpaces application using model driven design methodologies and how JeeWiz helped automate that process.

Last but not least is a session by one of our first partners, Fast Connect, who has done an amazing job in building some of the more advanced and demanding applications for leading French investment banks. These applications include advanced features such as WAN support, Compute/Data Grid integration, C++ and .Net Interoperability. In addition to all that, they also have to support very demanding performance requirements such as low latency, high throughput and extreme reliability. The Q&A portion of this presentation will be given by Christophe Benoit, a Senior Technical Architect at Société Générale.

In order to broaden the discussion a bit, we have included a session by Massimo Pazzini from Gartner. Massimo will speak about trends in the middleware market and how they shape up toward next generation infrastructure. Adrian Colyer from SpringSource, who has always been one of my favorite presenters, will also take part in the event, joining me in a panel discussion, which will be moderated by my dear friend John Davies.

Now you didn't expect that will leave the floor totally open, did you? Indeed, in this event, Yaron Benvenisti, CEO at GigaSpaces,  will make some really interesting announcements related to the future direction of GigaSpaces within the middleware and application-server market. Additionally, Uri Cohen, Product Manager at GigaSpaces, will speak about the new developments in our 6.5 and 6.6 releases. (We have notes on the new releases online, of course). Hopefully, we will show a pretty cool demo that will demonstrate how we deal with complex issues in building mission critical applications, such as reliability, scalability and performance, in a way simpler than ever. Specifically we are going to take a standard application and scale it in a matter of minutes. We will show how we can easily deploy it on our SLA-driven container and through that add self-healing and dynamic scaling capabilities to it. We will also show how we can remove a dependency on the database without sacrificing our application reliability all in matter of minutes using our new cloud tools package!

All in all, this event promises to be fairly exciting with lots of interesting stories, case-studies and updates. Having said all that, the true value of any event is mingling and networking - and this one is no exception. I'm looking forward to meeting many of our users, colleagues and others from the industry who have already taken part or want to become part of the new creation that we are building together. We will be setting up a room for one on one meetings for those interested in hearing more about potential use of our product, provide specific feedback around things they would like to see in future road-map or anything else that comes to mind. As seats are limited I'd recommend that you would register and ensure your participation. See registration details here.