Fitech Laboratories Inc. is pleased to announce the immediate availability of xTier/GRID 2.2 for Java – High Performance Grid Computing platform. xTier/GRID provides the API and infrastructure to execute tasks from one or more applications in parallel fashion across a grid of computers.
Almost any J2EE application that has any significant computational requirement can benefit from xTier/GRID. The power of grid computing can be implemented in your application in 3 simple steps: implement the task logic, use the task logic to implement a grid task, then run the grid task on the grid.
Check out our short online example with complete demonstration code, along with results from a 30 node test run that shows the simplicity and near linear scalability of xTier/GRID.
xTier can be deployed in homogeneous configurations or in a mixture of differing operating systems and hardware configurations. Essentially xTier/GRID runs anywhere there is a 1.4 JVM.
The key to the power of xTier/GRID for the J2EE application developer is that the feature set has been tailored to provide the developer with the right set of tools for implementing a robust grid enabled application without the unnecessary feature bloat of many existing grid products. xTier/GRID imposes very little overhead and has near linear scalability as compute nodes are added to the grid.
Read more about the grid service.
Download the xTier/GRID technical whitepaper.
Download a fully functional evaluation version of xTier/GRID.
-
xTier/GRID 2.2: Java-based Grid Computing Simplified (4 messages)
- Posted by: Jon Webster
- Posted on: December 16 2004 11:18 EST
Threaded Messages (4)
- xTier/GRID 2.2: Java-based Grid Computing Simplified by Brian Miller on December 21 2004 00:22 EST
- xTier/GRID 2.2: Java-based Grid Computing Simplified by Nikita Ivanov on December 21 2004 01:47 EST
-
Grid Computing Simplified by Maris Orbidans on December 21 2004 06:37 EST
- Grid Computing Simplified by Nikita Ivanov on December 21 2004 02:00 EST
-
Grid Computing Simplified by Maris Orbidans on December 21 2004 06:37 EST
- xTier/GRID 2.2: Java-based Grid Computing Simplified by Nikita Ivanov on December 21 2004 01:47 EST
-
xTier/GRID 2.2: Java-based Grid Computing Simplified[ Go to top ]
- Posted by: Brian Miller
- Posted on: December 21 2004 00:22 EST
- in response to Jon Webster
I see that both GridMathematica and xTier are dedicated to real-time interactive response from a cluster or desktop grid. I doubt the white paper's supposition that a WAN could dissipate a compute spike. I wonder why xTier doesn't use JXTA, which seems the best WAN fabric for tunneling, discovery, and quick installation. I wonder how xTier does these topology issues. -
xTier/GRID 2.2: Java-based Grid Computing Simplified[ Go to top ]
- Posted by: Nikita Ivanov
- Posted on: December 21 2004 01:47 EST
- in response to Brian Miller
Hi Brian,
I'm not entirely sure what you mean by "WAN ... dissipate a compute spike". But, essentially, xTier/GRID is for enterprise grids and we really don’t concern with WAN issues because global grids (and thus WAN-orientation) have very minimal business applicability due to number of reasons.
xTier/GRID has state of the art cluster capabilities including topology change auto-detection, pluggable failure resolution, virtual sub-clusters support, etc. xTier/GRID relies on that functionality to provide higher-level grid topology resolution.
As a side note, xTier is based on Service Oriented Approach, allowing clean separation between services yet maintaining conceptual integrity of the entire product. Check out our Javadoc and you’ll see what I am talking about.
Best,
Nikita. -
Grid Computing Simplified[ Go to top ]
- Posted by: Maris Orbidans
- Posted on: December 21 2004 06:37 EST
- in response to Nikita Ivanov
1) I think that grid computing is possible with clustered EJB servers, as well. Is xTier™/GRID much different from what can be achieved with clustered EJB's ? My first impression is that it may seem simpler just to deploy my session bean on several nodes, rather that writing tasks for xTier. Let's say, just because, many developers are already familiar with session beans.
2) I have done a small research about load balancing in clustered environments. http://maris.site.lv/loadbalancing/
Does xTier™/GRID do something to optimize requests in heterogeneous cluster - where nodes don't have the same hardware and/or operating system etc.
Maris -
Grid Computing Simplified[ Go to top ]
- Posted by: Nikita Ivanov
- Posted on: December 21 2004 14:00 EST
- in response to Maris Orbidans
Hi Maris,I think that grid computing is possible with clustered EJB servers, as well. Is xTier/GRID much different from what can be achieved with clustered EJB's ? My first impression is that it may seem simpler just to deploy my session bean on several nodes, rather that writing tasks for xTier. Let's say, just because, many developers are already familiar with session beans.
Generally, grid computing (and not just xTier/GRID specifically) provides substantially more than just clustered deployment of a resource. I don’t want to go into deep explanation here but the comparison is similar to saying "... I can just open TCP/IP socket and provide service instead of using app server with J2EE”. For anything but simplest cases it won’t work very well.
However, familiarity with J2EE is an important consideration and we fully address it.Does xTier/GRID do something to optimize requests in heterogeneous cluster - where nodes don't have the same hardware and/or operating system etc.
xTier/GRID was specifically designed to address heterogeneous grids. In fact, I believe we have most direct and comprehensive support for heterogeneous environment. We directly support grid taxonomy (allowing categorizing grid nodes by their performance characteristics), pluggable topology resolution and task routing.
Check out our example and other examples we ship with evaluation and see how our explicit split & aggregation design elegantly provides for optimal distribution over the heterogeneous grid yet allowing the developer to maintain the full control.
Regards,
Nikita.