Grid Meter » October 2006

October 31, 2006 | Comments: (0)

3tera's Grid Sandwich

TAGS: None

In a recent article Derrick Harris builds on the ever present theme of getting more visibility for Grid applications. The article begins with "Among the reasons often cited for enterprises not adopting Grid computing is a lack of Grid-enabled applications coming from the ISV community". Derrick cites the example of Callidus, a company who is indeed producing a Grid enabled application.

I've received some great responses to my own Grid Sandwich article in GRIDtoday. The underlying theme is that we need to start thinking from the application layer down at the same time as the network layer up when building Grids. One specific application I call out is the Apache Web server, which I believe would be pretty interesting on a Grid both for load balancing and redundant distribution of the data being served.

It turns out there is a company doing just that, 3tera", and they took the Grid sandwich approach packaging a concise solution both from the application on down AND the network on up.

3tera touts themselves as "the first grid operating system that runs and scales existing web applications." They approached this by creating a framework that encapsulates key applications like Apache and MySQL, among others, which are elements of the foundation for web applications. This framework, called AppLogic, has a unique GUI interface for tying web application building blocks together to create a web application Grid. So if you've built a web app, scalability, redundancy, fault tolerance, etc... all the wonderful promises of Grid are just a few "drags and drops" away.

Now the catch is you have to play on their network, or at least the network of their partner, Utility Serve, so if you have a room full of hardware and are looking for an in-house solution, you will have to convince 3tera to come and install their Grid OS in your house, which they have indeed done for others. The neat thing about their managed hosting solution is that the Grid OS component is billed based on memory usage. Which reminds me, Service Level Agreements and billing strategy discussions for Grid are something I'd like to see a little more talk about in the near future.

But what about non-web based applications? As long as these applications are Linux based and there are no hardware dependencies, the folks at 3tera assure me that it is a straight forward process to get them up and running as AppLogic apps and that they are standing by to help.

Posted by Greg Nawrocki on October 31, 2006 08:57 AM

October 25, 2006 | Comments: (0)

Embedded Grid in the Press

TAGS: None

Here's a recent article that touches on my favorite topic of the use of Grid computing in embedded systems.

Granted, the article is a little light, so I'm not sure if these are just sensors that talk to each other or true Grid nodes in the "three point checklist" sense. And I'm also not so sure that "a river that floods regularly after Christmas every year," is the right application for a collection of wireless sensors scattered about or would something a bit more permanent be a better answer. The use of wireless networking when wired should be used is also a bit of a pet peeve of mine.

Nevertheless, it is a bit of press that raises the interest in a market (embedded devices) that I believe is an area where we will see the true promise of Grid shine through.

Posted by Greg Nawrocki on October 25, 2006 08:35 AM

October 24, 2006 | Comments: (0)

Platform Computing and "Virtualocity"

TAGS: None

Platform Computing has always been an interesting company to watch. They have been part of the Grid game since before the words "Grid" and "Computing" were used together, and they have supported themselves entirely through customer sales with no money from venture funding.

Platform CEO Songnian Zhou is adept at keeping the driving forces behind the company fresh while staying true to the ideals that have been the foundation of the company since its inception. I believe this is illustrated well in this article in IT Week.

Zhou uses the term virtualization to explain how the focus of Platform's EGO software is to "help firms orchestrate and manage grid services" as CIOs turn their focus from managing a collection of individual boxes to managing the datacenter as a virtual environment. This is very similar to what he said way back in January in this interview with the Globus Consortium Journal. "The killer app for grid is not the components of applications - ERP, CRM - but the connection of all of these components of enterprise applications to form business processes." In other words, it's not the individual applications but their aggregation in a virtual environment that unleashes their true power, that of driving business processes.

As Zhou states "Even the grid buzzwords are short for other buzzwords so if we focus on the business need than the rest doesn't matter." If Platform wants virtualization to mean dynamic system management to speed business processes then I say more power to them. And if it inspires others to use made up words like "virtualocity", even better.

Posted by Greg Nawrocki on October 24, 2006 09:30 AM

October 18, 2006 | Comments: (0)

Grid Applications, Getting Closer?

TAGS: None

My recent Grid sandwich article in GRIDtoday has me all fired up again about Grid and applications. After a quick scan of recent Grid news this morning I saw this article from a Sun executive that had mention of both their Star Office suite and their efforts in Grid computing.

Well we're getting closer, we've got Grid and Star Office mentioned in the same article, the trick now is to get them mentioned in the same paragraph...

Posted by Greg Nawrocki on October 18, 2006 10:04 AM

October 16, 2006 | Comments: (0)

Grid Economics and Business Models

TAGS: None

The Journal of Grid Computing is set to publish a special issue on Grid Economics and Business Models. This was mentioned today in GRIDtoday and the call for papers is available here in .pdf format.

Economics is a pet interest of mine, and application of this science to the science of Grid is something that analysts attempt to do all the time. With this special issue, anyone who responds to the call for papers gets to be an analyst. This is sure to provide some fascinating and hopefully inspiring results.

Given the usual scientific nature of this publication one might be concerned that it may yield more academic discussions as opposed to those more suited to the business practical. However, I'll point out that the special issue co-editor is Wolfgang Gentzsch. While Dr. Gentzsch carries with him impressive academic credentials to be sure, he also has a wealth of enterprise experience, notably with Sun Microsystems, and has first hand experience with large scale grid deployments.

This special issue is sure to be a "must read".

Posted by Greg Nawrocki on October 16, 2006 09:18 AM

October 09, 2006 | Comments: (0)

Grid Applications and Contests

TAGS: None

I recently reconnected with an old friend of mine. I've known this person since we were both four years old and we kind of lost touch sometime around the high school years when he moved away. Anyhow, it turns out that he is now working for the FBI and I'd like to think that our childhood games of "cops and robbers" had some influence on his career choice. Even with all of the tools that I'm sure are available to him for searching out people, he simply plugged my name into Google and the breadcrumbs were easy to follow.

After he told me how I was discovered I got to thinking. I had honestly never done a Google search on myself. I mean why would I, I almost always know where I am. I gave it a go anyhow and out popped pretty much what I was suspecting. However, as I dug a little deeper I found items from my long ago (or at least it seems long ago) past such as documentation on particle accelerator control systems that I had authored. OK, so none of what I just said really has any bearing on the topic and hand, but bear with me...

I also happened upon a blog by Dan Hushon, Chief Technologist, Sun Grid. In his blog Dan calls me out for a couple of my past blogs where I question whether grid computing contests the right driver for enterprise, and I suspect one in particular where I am less than supportive of the Sun Cool Apps Developer Contest. But it got me to thinking, what every happened to that contest!

Well, sure enough the contest did have an end and there were winners. Among the winners were an application that grows an optimal neural network of variable size to maximize memory capacity and minimize network size for any training set, a framework and plug-in for the NetBeans IDE, and a general purpose indexing engine and a 3D Fractal Rendering project.

Now I certainly don't mean to belittle any of these projects or the contest winners. I'm quite sure they are each smarter than I can ever hope to be and I shudder to think what my coding skills would have come up with if I blew the cobwebs off. However, none of the applications mentioned seem all that mainstream to me, and some don't sound like applications at all, but enablers for applications.

In Dan's blog he asks if I know of a better way to get to our end goal of agile-distributed-component based-failure resistant-capacity managed-systems. My suggestion is to work from the other side of the problem. Take a known application or application type that suffers from the "never big enough, never fast enough" resource problem Dan mentions in his blog. Again I'll harp on the "Microsoft Excel or equivalent" spread sheet as a great place to start. Instead of building a system that applications may or may not ever be built for, start with a known application and build the system around it.

Posted by Greg Nawrocki on October 9, 2006 01:12 PM

October 05, 2006 | Comments: (0)

Taking a Read of the "Grid Gauge"

TAGS: None

There was an interesting "Ask Slashdot" posting recently on Slashdot. In a nutshell, the person posting the question works for a large investment bank and was being pushed to use Grid technology. The question this person was posing was "To Grid or Not To Grid?".

In typical Slashdot fashion, especially as it pertains to the Ask Slashdot category, the responses were varied, some tongue in cheek, and application of intelligent filtering always advisable.

Two of the things that did strike me about the responses were 1) There is still a confusion about exactly what "Grid" means; and 2) that there were very few "Grid is a pipe dream" comments, but instead the emphasis was along the lines of "Grid is real, but it ain't easy". The reason that these points resonated with me is that I just recently completed compiling my survey from GlobusWORLD / GridWorld 2006 and that my data was showing some of the same trends.

In my past two years of surveys I have asked the question "Do you feel there is confusion regarding the term 'Grid computing'?" Over the last two years those responding to my surveys have stayed pretty much consistent at 60% saying "yes" and 40% saying "no". The stated concerns echoed the Slashdot article and had mainly to do with the confusion between what is Grid and what is Clustering and confusion between related terms such as Grid and Virtualization.

I've also noticed a trend over the past year that my survey respondents have become less concerned with issues of software stability and increasingly concerned with the difficulty of implementing and administrating a Grid solution.

I'm certainly not going to defend my read of Slashdot as statistical science, however I believe we are indeed seeing the first steps toward mainstream acceptance of Grid. While we may not all agree what "Grid" means, we aren't so much concerned if it will work, but how straight forward it is going to be to get it deployed.

Posted by Greg Nawrocki on October 5, 2006 12:39 PM

October 02, 2006 | Comments: (0)

A SMB Grid Product to Watch

TAGS: None

One of the things that draws so much fire from critics, including myself, is that the generic term "Grid" means so many things to so many different people. The positive flip side to this is that the Grid door remains open to products that target a variety of different markets for a variety of different applications.

In the early days of Grid we often touted the ability to create aggregates of existing equipment to accomplish new tasks, the resulting total greater than the sum of the parts. Lately we seem to hear more from those targeting big enterprise computing who are generally more concerned with getting all they can from those new blade servers they purchased. While the price point of new equipment is certainly within reach of small to medium size businesses (SMBs) these folks often simply do not have the manpower to implement a new system and need products that work with their current network and storage hardware.

I recently stumbled upon a company targeting the SMB market, RevStor. RevStor has a product in the pre beta stage called SWARMS. SWARMS is an interesting product that builds a turn key data Grid on a variety of operating systems running on commodity hardware. I've seen it demoed and it quite simply just plain works.

I'm looking forward to keeping tabs on RevStor. Not only because I think they have a nice concise workable product, but I believe that they will be a litmus test for main stream Grid acceptance in the SMB market.

Posted by Greg Nawrocki on October 2, 2006 09:57 AM