2004 was a busy year for enterprise grid computing.
We saw a dramatic increase in the volume of industry discussion and media and analyst attention around commercial-level deployment issues. We heard announcements from major vendors about large enterprise initiatives. And we saw some important discussions start to unfold around practical issues related to grid adoption on the enterprise scale -- issues such as how to best secure a distributed grid environment (see story), as well as software licensing issues (see story).
As we look ahead to 2005, I thought it would be interesting to catch up with some of the leaders from the grid industry and have them weigh in on what they consider to be important issues to watch in the coming year. Responses varied from specific concerns to philosophical perspectives about the implications of grid technology.
Here's what they had to say:
"First, we believe that the concepts of service-oriented architectures (SOA), utility computing and data center automation will continue to gain traction through 2005. Grid computing should benefit exponentially from this when it is integrated with Web services and as its use as a means to underpin these activities becomes self-evident.
"Second, the key challenges through 2005 will be enabling data management on grids, bringing commercial applications to grids, developing integrated infrastructure for enterprise grids and using grids to support SOA. There also is a significant need for new approaches to software licensing for grids, which will become an overriding factor and could hinder wider commercial adoption of grids if this issue is not addressed.
"Finally, some things to ponder: Where are the application-server vendors? And where is Microsoft? A lot to think about going into 2005, but also a lot of opportunity and activity, which is good."
-- William Fellows, principal analyst, The 451 Group
"It's time for IT to share computing cross-departmentally, rather than clinging to resources like they're personal possessions. It's interesting how people don't worry about whose electrons they get from the power company or whose raindrops they get from the faucet -- but how IT professionals are obsessed with where they get their computing power. There seems to be a strong affiliation today with the notion of 'I need computing from a certain machine.' A good New Year's resolution for IT professionals would be to drop this limited view of IT resource consumption and truly embrace the reality that distributed resource sharing is the future of IT."
-- Rich Green, executive vice president of products, Cassatt Corp.
"My wish is that the production-proven, publicly referenceable business benefits of grid computing finally overshadow the hype surrounding this rapidly emerging technology. ... Grid computing delivers real business benefits and cost savings for the 'virtual' enterprise."
-- Peter Lee, CEO, DataSynapse Inc.
"Grid and its associated community of developers, implementers and users may see 2005 as a seminal year. I expect demonstrable progress will be made in the area of standards. It is crucial that grids be based on open standards to ensure interoperability and to future-proof a CIO's grid investment. Standards such as WS-RF [WS-Resource Framework], WSDM [Web Services Distributed Management], The Globus Alliance and the Global Grid Forum will arrive in specifications and reference implementations. ... I am also hopeful that 2005 will yield grid implementations in Fortune 1,000 companies in mission-critical business or transactional applications."
-- Greg Astfalk, chief scientist, Hewlett-Packard Co.
"Barriers to grid adoption in the enterprise are becoming less of an obstacle as the technology matures and customers continue to realize significant business value from their grid implementations. The largest barrier in 2005 is the adoption of standards by commercial providers of technology.
"Most enterprise customers have very complex heterogeneous IT environments. In order to achieve accelerated time-to-value from grid solutions, the implementation of grid software technologies must be painless. Adoption of open standards built on core Web services standards (such as WS-Addressing, WS-RF and WS-Notification) by commercial IT providers of technology significantly reduces the complexity of grid implementations in the enterprise. As the complexity goes down, customers will achieve greater business value and faster ROI from grid solutions, especially as they move from line-of-business to enterprisewide implementations."
-- Ken King, vice president of grid computing, IBM
"It's been a long journey, but the ability to put together a 'virtual data center' which can support real-world applications is almost here. Today's challenge revolves more around getting the available technology to support today's real applications than the capability of the basic technology.
"Each of the major suppliers -- HP, IBM, Microsoft, Oracle, Sun and a few others -- has put forward a broad strategy. Many smaller companies, such as DataSynapse, Cassatt, Meiosys, OpsForce, Qlusters and others, are offering innovative solutions for a certain set of application requirements.
"As the standards evolve and are adopted by the suppliers, organizations will be able to deploy their preferred tools without facing barriers and incompatibilities. When that occurs, the virtual data center will be a reality."
-- Dan Kusnetzky, analyst, IDC
"The most significant adoption barrier to grid in mainstream enterprise environments is the broad availability of standards-based solutions and building blocks that provide the flexibility and interoperability necessary to integrate all aspects of the enterprise, from the wireless edge to the data center back end.
"A service-oriented approach is proving to be the architecture of choice. ... Services should also be used to integrate data sources as well as the supply and deliver chains of a business."
-- Robert Fogel, director of grid strategy and business development, Intel Corp.
"I have some concerns about the continuous political tiffs that sometimes distract attention from implementing real, working grid infrastructures, including debates over the roles and contributions of the different standards organizations, fundamental definitions and broad strategic directions.
"However, I envision 2005 as the year with a dramatically growing number of real grid implementations in research and industry based on a growing consensus in the community about a common service-oriented architecture and Globus Toolkit 4 as one of its reference implementations, agreement on open standards and their wider adoption, continued funding of advanced grid projects from the governments and large grid testbeds with more tangible results."
-- Wolfgang Gentzsch, managing director, Microelectronics Center of North Carolina Grid Computing & Networking Services
"In Nemertes' benchmark research 'Extending the Enterprise,' we asked IT executives whether they used grid technologies. Only 8% of participants had deployed some form of grid computing, and they were concentrated in two industries: financial services and pharmaceuticals. The same group of participants reported that 75% had deployed some Web services based on XML and SOAP [Simple Object Access Protocol].
"Our analysis suggests that grid computing and Web services in the form of an SOA are feeding each other's growth. Greater adoption of SOAs makes software applications more modular and easier to deploy on a grid infrastructure. Meanwhile, grid computing increases infrastructure utilization and flexibility while lowering operational costs.
"IT executives deploy grid computing to increase business agility and reduce costs and then develop more SOA applications to run on their grid. The greatest obstacle to grid adoption is, therefore, the cost of recoding enterprise applications. In the long run, we believe that the cost savings and flexibility created by adopting grid computing and SOA will far outweigh the costs."
-- Andreas Antonopoulos, principal analyst, Nemertes Research LLC
"The Wright brothers had plenty of thrust to go airborne. It is control that was highly problematic. In 2005, the organizations that implement medium-to-large-scale grids will face a similar challenge. While resources abound, the virtualization and dependable orchestration of the same will prove quite brittle still. Even more so once new resource types -- networks, devices, sensors -- surge to the status of grid-managed resources.
"I wish to have a 2005 punctuated by intra- and extragrid proof points at [Fortune 100 companies], with the 'big science' projects continuing strong as beacons for grid innovation."
-- Franco Travostino, director of Advanced Technology, Nortel Networks
"There is a lot of hype with regard to grid computing, and this hype will continue and maybe increase in 2005. In my opinion, it is essential in this phase to stay focused, meaning quality is more important than quantity (e.g., fewer grid working groups with a higher quality level, fewer tools -- but thoroughly tested ones that are usable on all platforms, etc.). Staying focused also means doing something to gain experience, which is sometimes faster and more valuable than talking about what could be done. So my wish and goal for 2005 is to stay focused to deliver real value."
-- Alexander Gebhart, development manager for NetWeaver enterprise grid computing, SAP AG
"Grid is at the tipping point of becoming mainstream in enterprise computing. To make this transition, we need to deliver a network computing architecture that treats applications as services deployed on a dynamically scalable, configurable and self-healing resource fabric. Most importantly, these capabilities need to be delivered in an open, robust and scalable environment that can perform against contracted service levels. This is a nontrivial task that takes a community. Sun is a strong supporter of the community process that is creating the specifications, standards and technologies required to accelerate the adoption of grid computing in the enterprise."
-- Peter ffoulkes, director of grid marketing, Sun Microsystems Inc.
"As we learned from the Internet's progression from academia to business and Linux's leap from open-source community project to mainstream operating system, open standards and professional support are two critical pillars for the commercial success of an emerging technology. Univa believes that 2005 will see the emergence of grid standards supported by a critical mass of technology providers.
"We also expect that implementations of these standards based on commercially supported open-source software, such as the Globus Toolkit, will provide the necessary proof points for broader adoption of grid technologies by enterprises. These combined trends will transform grid from academic curiosity to business enabler and lay the foundation for rapid growth of the technology in 2006 and beyond."
-- Steve Tuecke, CEO, Univa Corp.
Ian Foster is a co-founder of The Globus Alliance and an advocate of open-source grid. He can be reached at foster@mcs.anl.gov.