A short interview in Network Magazine India with Gregory Corgan, Executive Vice President, Worldwide Sales, Computer Associates. He gives their perspective on taking their grid services to the business community in India.

"Witnessing an Evolution" is an intro article on the Government Technology magazine's site. Two US state projects are mentioned, the San Diego Supercomputer Center's work on making terabytes of survey and map data accessible, and the Global Grid Exchange in West Virginia which aims to offer utility support for business in the state.

The article suggests the term "the grid" has "inherent Orwellian overtones", but perhaps some things are just in the eye of the beholder…

The word "grid" is heavily overloaded, and ActiveGrid's usage may not please the pedant, at first sight they're talking about clustering. However, on closer examination their ActiveGrid Application Server does have interesting characteristics, and in some respects does crossover into the purist's definition.

Their architecture is built on the popular open source LAMP stack (Linux, Apache, MySQL, Perl/Python/PHP)  on commodity computers, but a significant addition is their use of XML languages such as BPEL Business Process Execution Language) to define application flow. Individual services may be coded in virtually any language you like, but the wiring between them is defined in  declarative XML.

ActiveGrid describe themselves as a "commercial open source company", and among their downloads there is a rapid application development tool (various platforms) along with an Early Access release of the application server itself. This InfoWorld article by Jon Udell gives a good overview, and there's an (audio) interview with Peter Yared of ActiveGrid at IT Conversations.

According to this Information Week article, MySQL AB is partnering with Red Hat Inc. to support users running the database on Linux as companies build larger server clusters and grids.  This seems to be "grid" in the lower-case sense of clusters. It's an interesting development considering RedHat previously favoured Postgresql as their database of choice, even releasing their own version.

But MySQL has made significant advances in recent years in terms of replication capability (see the slides from Jeremy Zawodny of Yahoo on the issue). That Linux plus MySQL may be a good foundation for more upper-case Grid systems  (when you add Apache and one of the P* languages, together with a sprinkle of XML) is suggested by the ActiveGrid Grid Application Server.

In a nearby space Adam Bosworth, one of the Google guys who has the sleeves of his lab coat rolled up has also been talking up MySQL as a possible solution to distributed data on the Web. Bosworth also has the rather controversial suggestion that the content distribution capabilities offered by RSS 2.0 and Atom with their "sloppy" approach to extensibility may provide the necessary inter-system protocols for moving data around.

The release date of the long-awaited Globus Toolkit 4.0 (GT4) final is April 29th.  This is the leading open grid tool set, and the new release will be considerably more aligned to Web Services than previously (all GT4 Web service interfaces will be WS-I Basic Profile compliant), so expect a burst of Web/Grid activity.

ClusterWorld Summit

  focusing on clustering solutions for business, research and academia. 
Millbrae, California, May 16-18, 2005. Organised by ClusterWorld magazine.

The processing for CERN's Large Hadron Collider will be handled by an internationally distributed grid, which will be the world's largest. It's due to start full operation in 2007, simulating the conditions of the universe's first moments. An incongruously nationalistic post on the CeBIT blog, British effort builds world's biggest grid quotes project leader Les Robertson as saying:

"We are well ahead of our original schedule for reaching 100 sites…Thanks is due to the many partner sites around the world for their contribution to this success. Making a grid like this is a truly collaborative effort."

India-based Satyam Computer Services Ltd. have announced (press release) a strategic alliance with US-based Grid specialists United Devices. The strategy seems quite clear - UD have the technologies and some level of market coverage in the West, Satyam have already piloted Grids and have widespread market coverage for managed solutions in the East. The combination is seriously global.

The OASIS WSDM TC have decided to make the Web Services Distributed Management specifications an OASIS standard.

However, it seems that the documents  have as normative references other unfinished specs, for example referring to previous versions of the WSRF specs. There's been considerable concern in the blogosphere, where the popular feeling is that not only is the WSDM spec questionable (and driven by proprietary goals), but it reflects very badly on OASIS as a whole. Commentary: Greg Pavlik, Savas Parastatidis, Tim Bray.

pyGridWare

is a Python implementation of the WS-Resource framework. It includes support for WS-Addressing, WS-Notification, WS-Lifetime management, and WS-Security.

Project goals include interoperability with the Globus Toolkit4 and support for GSI Secure Conversation, Secure Messaging, and XML Signatures.

"...ever been driving a tank in a game like Call of Duty: United Offensive and been stopped by a shrub?"

. There's a solution on the way.

The PhysX(TM) chip from AGEIA is is a "Physics Processing Unit" (PPU) targetted at gamers. It takes care of the number crunching needed to simulate interactions in the physical world. This gamers-depot article suggests the current limit for game-oriented modeling is 30-40 bodies on today's high-end CPUs. The new chip will take that to a maximum of 40,000.

AGEIA hope to have cards for standard PCs available by the end of the year.

Press Release.

is a CEO/programmer in the biotech industry based in Amsterdam. He's also a blogger who regularly posts on Grid-related topics (Atom feed).
Here's his latest batch of relevant links:

Grid Information (at SARA Computing and Networking Services)

Rocks Clusters (the "Award Winning Open Source High Performance Linux Cluster Solution")

cfengine - Configuration engine

OpenPBS - The old and "reliable" PBS.

Torque - A newer queueing system

Grid Howto for the LHC (Large Hadron Collider) at CERN in Switzerland.

Batch processing on Lisa - it uses Torque as the scheduling framework and Maui as the scheduler.

No doubt in response to Sun's Grid-for-hire, IBM have announced a pay-as-you-go service on their Blue Gene system $10,000 per week will give you access to one-eighth of rack, which amounts to 128 dual-core processors. On the other hand, you might consider buying a rack outright - if you have $2 million handy.

See also: overview article at Red Herring.

Hardware time - over at Tweakers.net there's a thorough review of nine Serial ATA RAID 5 adapters. It discusses the architecture of the various RAID implementations and gives a detailed view on the performance of nine of the most popular SATA RAID solutions.

Overall conclusion in brief: SATA RAID is generally comparable to its SCSI counterpart, and ready for the enterprise.

(spotter: ClusterWorld)

Fujitsu Siemens Computers have announced their IT infrastructure solution FlexFrame now has Solaris operating system (and PRIMEPOWER hardware)support. FlexFrame for mySAP Business Suite is a modular system aimed at helping consolidating mySAP solutions on a single easily managed platform. It allows hardware resources to be allocated on demand.