Grid Meter » February 2006

February 28, 2006 | Comments: (0)

Government Security Concerns Bog Down Sun's Grid Deployment

TAGS: None

According to a Bob McMillan article today, the U.S. government is stifling Sun's Grid deployments.

The lead quote by Jonathan Schwartz sounds the trailer for a bad action flick, but illustrates cultural hurdles that need to be overcome before Grid has a wide acceptance. This time however the perspective is from the outside looking in.

"Our servers are considered munitions by the federal government," Schwartz said. "So when we wanted to provision servers off of our computers to a global population, the federal government got involved and said, 'we'd like to know all the people who use this.'"

Sun Grid seems to be getting the double whammy and is being hit from the inside out as well in this battle of changing hearts and minds toward Grid deployment. The article points out an example of a potential customer that's started to get cold feet about doing things differently.

"Sun CEO Scott McNealy said that this kind of hesitation ultimately scuttled a Sun Grid deployment with an unnamed bank. 'The bank made so many security and compliance-related demands that Sun eventually walked away from the deal,' he said."

Sun's marketing and PR is generally pretty interesting but these particular events are a great example of what a knee-jerk topic security is, and how stubbornly a customer or big brother can tank a deployment with the "s" word.

Next week Grid Meter will be spotlighting some pretty compelling Grid security efforts underway today (in both commercial and research / science settings). If you have an interesting Grid security story to tell, drop me a note.

Posted by Greg Nawrocki on February 28, 2006 09:03 AM

February 22, 2006 | Comments: (0)

today's SOA developers .... tomorrow's Grid developers

TAGS: None

Three years ago, Grid computing and SOA were both commonly regarded as "nascent, emerging enterprise technologies."

But while enterprise Grid uptake is still at its infancy, service-oriented architecture interest has blown up in the commercial sector. Go to InfoWorld's SOA Executive Forum, and you will hear very mainstream enterprise end users talking about real world deployment issues and experiences. Go to GridWorld, and in the enterprise track, you'll hear from a smattering of early adopters -- and then a lot of vendors describing what the future will look like.

Why has SOA jumped out ahead of Grid in enterprise traction?

There are a range of factors that have supported the explosion of commercial interest in SOAs. Web services standards have matured tremendously over the last five years. Developers have a wide range of options to choose from in programming languages -- not just Java, but .Net, Ruby on Rails, the "P" languages (Perl, Python and PHP). But I think the most compelling reason why SOA has outpaced Grid in enterprise growth is that you can actually take relatively mainstream enterprise applications and write to a service with a simple API. That just hasn't happened yet in Grid.

But one really encouraging fact is that the common web services skills and knowledge that SOA is ushering into the enterprise are extremely similar to the skills required to write Grid services. Last year's release of Globus Toolkit v 4.0 signified the Grid community's full embrace of the WSRF. If you poke around today's technical tutorials on building Grids -- such as the recent IBM developerWorks series on "Building a Grid using Web services standards" -- you'll note that SOA and Grid architecture development practices have truly converged.

So if you're keeping an eye towards the future, and you want to be in a good position for your company to take advantage of Grid when the technology has fully matured -- just keep plugging away on SOA, and you'll be fine.

Posted by Greg Nawrocki on February 22, 2006 07:21 AM

February 21, 2006 | Comments: (0)

Manufacturing Industry Prime for Data Grids

TAGS: None

Recently, members of the Grid community (myself included) have been harping on the fact that it's data management -- not just compute horsepower -- that's accelerating enterprise interest in Grid.

And manufacturing will be a particularly important vertical market to watch in enterprise Grid uptake. Here's an industry that needs to leverage both compute and data-intensive processes. The classic example is automotive clash analysis.

When designing something with complex interacting systems -- like an airplane or automobile -- the smallest change in one part can affect many others, including parts in other systems. An engine that can't fit between the front fenders come installation time is less than practical. In those complex designs, when a change is made, massive simulations are run to aggregate data from many different systems and ensure that the laws of physics are being met, and that two parts are not occupying the same space (hence avoiding the proverbial "clash").

In a recent announcement by Grid vendor Altair -- which has always had its sights trained on the manufacturing vertical -- they similarly echo the importance of data management in the Grid discussion:

"In his introductory editorial, James R. Scapa, Altair president and CEO, discusses corporate information and historical data, which he believes are increasingly valuable assets to a growing number of industries. Further, Scapa sees that organizing, optimizing and visualizing performance data is essential in providing the insightful information needed to make sound business decisions."

Altair is the company that brought to market the (very widely used) PBS / PBS Pro products -- synonymous with job scheduling on computational Grids. But it sounds like Altair is upping the ante on its data Grid story as well.

Posted by Greg Nawrocki on February 21, 2006 06:00 AM

February 20, 2006 | Comments: (0)

Manufacturing Industry Prime for Data Grids

TAGS: None

Recently, members of the Grid community (myself included) have been harping on the fact that it's data management -- not just compute horsepower -- that's accelerating enterprise interest in Grid.

And manufacturing will be a particularly important vertical market to watch in enterprise Grid uptake. Here's an industry that needs to leverage both compute and data-intensive processes. The classic example is automotive clash analysis.

When designing something with complex interacting systems -- like an airplane or automobile -- the smallest change in one part can affect many others, including parts in other systems. An engine that can't fit between the front fenders come installation time is less than practical. In those complex designs, when a change is made, massive simulations are run to aggregate data from many different systems and ensure that the laws of physics are being met, and that two parts are not occupying the same space (hence avoiding the proverbial "clash").

In a recent announcement by Grid vendor Altair -- which has always had its sights trained on the manufacturing vertical -- they similarly echo the importance of data management in the Grid discussion:

"In his introductory editorial, James R. Scapa, Altair president and CEO, discusses corporate information and historical data, which he believes are increasingly valuable assets to a growing number of industries. Further, Scapa sees that organizing, optimizing and visualizing performance data is essential in providing the insightful information needed to make sound business decisions."

Altair is the company that brought to market the (very widely used) PBS / PBS Pro products -- synonymous with job scheduling on computational Grids. But it sounds like Altair is upping the ante on its data Grid story as well.

Posted by Greg Nawrocki on February 20, 2006 07:00 AM

February 16, 2006 | Comments: (0)

Impressive Early Performance by CERN's LHC Grid

TAGS: None

Todd Weiss from Computerworld posted an interesting write-up today about impressive early data transfer results in CERN's Large Hadron Collider (LHC) Grid effort.

According to Weiss, "The 100,000-PC grid will help CERN and a worldwide network of researchers and scientists harness a huge amount of computing power and storage capacity that it couldn't build on its own. The grid will also be able to store some 15 petabytes of data -- 15 million gigabytes -- annually, and will have to be continually expanded because the data will have to be continuously preserved for analysis."

In other words -- that's a heck of a lot of data that's going to be transferred on an ongoing basis for the LHC. No doubt large enterprises that grapple with management of huge data sets should keep an eye on CERN's Grid efforts. After all, enterprise data management has more in common with research / science data management than people tend to realize.

Posted by Greg Nawrocki on February 16, 2006 12:48 PM

February 16, 2006 | Comments: (0)

EGA, GGF Merger in Perspective

TAGS: None

Roughly two years ago, Oracle kicked off the Enterprise Grid Alliance -- dedicated to the enterprise advancement of Grid computing. The EGA vowed: "With our pragmatic approach focusing on the enterprise, EGA is uniquely positioned to deliver near term, tangible benefits."

The EGA immediately took the PR posturing that only they truly understood the unique requirements of enterprise Grid ("This isn't some academic group that's trying to boil the ocean"). They made pointed efforts to polarize the Grid discussion -- researchers and scientists who just 'don't get' enterprise Grid on one side, and EGA paying members on the other.

The EGA's abrasive positioning drew some criticism from the core Grid computing community - as well as some comparisons to how commercial interests similarly tried to cast aside the research / academic pioneers as the Internet evolved into mainstream use.

But that was then and this is now...

We've heard rumblings for months now, however it seems like the goal of all joining hands and singing "Kumbaya" is ever closer to reality.

What's interesting about the EGA is that the lines it drew in the sand were truly artificial. The participating IT vendors (the roster is impressive) have all been actively participating with the GGF for years. Many have developed time / money / code to open source Grid efforts in e-Science (such as the Globus
Toolkit) -- and if you look at the largest deployed Grids today (such as Cern and TeraGrid), you'll see a wide representation of their equipment in use.

The GGF has always represented tangible results and progress -- a group of technologists truly working together on common goals for Grid. However, both organizations have often been criticized as to whether they have been ultimately focused on the top of the value chain, the end user. The two organizations were a bit like shopping carts with an errant wheel, a bit tough for the users to push down the aisle toward successful grid adoption and deployment.

The GGF has made great strides in the past year bringing commercial users, implementers and vendors into the fold. I attribute that to Mark Linesch's leadership and his ability to read the pulse of enterprise Grid.

With two good wheels for Mark to choose from, I'm expecting a cart that glides down the center of the aisle ...

Posted by Greg Nawrocki on February 16, 2006 11:52 AM

February 15, 2006 | Comments: (0)

"Grid Computing for Developers" hits the shelves ...

TAGS: None

For enterprise software engineers that want to cut their teeth on Grid computing development, a new release by Charles River Media, Inc. "Grid Computing for Developers" might be worth taking a look at.

Author Vladimir Silva is a former IBM software engineer with deep technical experience with the IBM Grid Toolbox and the Globus Toolkit. His book drills beyond conceptual Grid architecture discussions -- and provides a companion's guide (with code samples) for common Grid installation, security set-up and integration scenarios.

Silva notes that the enterprise is still in the early exploration days in terms of Grid-enabling core business applications -- but "Grid Computing for Developers" does provide some specific technical drill-downs on Grid services running in more 'mainstream' commercial types of environments. For example, in the "Security" section (Chapter 8), Silva describes how a certificate authority works with a Web-based application "along with command-line tools and installation transcripts for two popular application servers: Apache Tomcat and IBM WebSphere."

While the book went to publication prior to the release of GT4 (and therefore emphasizes best development practices around GT3), 'Grid Development for Developers' stands out as one of the first Grid publications that gives the enterprise developer a technical point of reference as they actually role up their sleeves and get their hands dirty with some of the core open source Grid plumbing in common use today in research and science Grids.

Posted by Greg Nawrocki on February 15, 2006 07:36 AM

February 08, 2006 | Comments: (0)

Podcast: Efforts to make Grid more relational data-friendly

TAGS: None

Six years ago, a growing group of Grid pros in the UK realized that a diverse range of data resources (xml collections, relational databases and other structured files) were not being accommodated very well in Grid environments. Grids in science have tended to deal with "blob data" (unstructured data) very well, but had trouble with relational data.

So the OGSA-DAI (Data Access and Integration Services) project was born. Today this project, based at the University of Edinburgh, is increasing the efficiency and reliability with which these data sources are being managed and accessed on the Grid.

Here's a podcast with Neil P. Chue Hong, one of the technical leaders of the project:

Posted by Greg Nawrocki on February 8, 2006 08:45 AM

February 06, 2006 | Comments: (0)

A Virtualization Pro's Take on Grid Challenges

TAGS: None

I recently had the opportunity to speak with Simon Crosby - CTO of XenSource, the Palo Alto, CA-based vendor that 'plays the dual role of leading the open source Xen [virtualization] community, while simultaneously selling value-added enterprise solutions based on Xen.'

Crosby cited Grid's problems handling relational data as a major challenge ...

"The Grid's I've seen in action don't deal with complex storage architectures -- the data sets are local to the computation or are made available from a centralized server. But for a broader class of enterprise applications, the storage problem must be addressed. Many applications are now interlinked ... for example, a Web service today is composed of a Web server, an app server, and a database."

... and had an interesting point about the difference between managing applications on the Grid, versus managing virtual machines on the Grid.

"We can instantly clone a running VM and distribute it across multiple servers in the datacenter. The Grid world understands how to use such features because the Grid community understands the concept of quickly creating multiple copies of an application on multiple systems. The key difference between scheduling an application to harvest resources from the Grid, and scheduling VMs onto computational and storage resources is that applications in the Grid world today tend to be of finite life-span (for example, consuming spare CPU to run a simulation) and will terminate at some point. But VMs are instances of operating systems, that essentially run forever until they are 'turned off.' VMs are more fundamental, perhaps more elemental than applications."

Posted by Greg Nawrocki on February 6, 2006 06:18 AM

February 01, 2006 | Comments: (0)

Unlocking Grid Security Complexity

TAGS: None

In enterprise environments, it is rather trivial to get web servers, scripting languages, databases and common systems up and running. In many OS distributions, these systems are a check box option in a simple admin GUI.

Not so with Grid computing, where intricate security dependencies can really complicate things.

Here's a link to an IBM developerWorks tutorial that highlights some of these complex dependencies and discusses a unique approach, the Grid Accounts Management Architecture (GAMA) to unify Grid security components and present a more concise interface for Grid deployment. Tools like GAMA are bringing the goal of Grid ubiquity closer to reality. The entire set of articles in this "Building a Unified Grid" series is a great reference for enterprise developers building an end-to-end Grid process.

Posted by Greg Nawrocki on February 1, 2006 07:55 AM