Home · About · A-Z Index · Search · Contacts · Press · Register · Login
|
|
|
Home · About · A-Z Index · Search · Contacts · Press · Register · Login |
|
Plenary - Boundaryless
Information Flow:
|
Beyond the Hope and the Hype: Customer Perspectives on Achieving the Vision of Grid Computing |
 |
Professor
Bill Estrem
|
Bill explained that the Customer Council has conducted a survey of customer members, asking them to respond informally about their perceptions on the impact of Grid computing using four questions:
Bill said that listening to the speakers in Monday's Plenary, we all recognize there are risks and barriers we need to address. Responses to our survey comprised a relatively small number representing very large organizations, with very few from small and medium enterprises. Respondents were enterprise architects from large organizations across several industry verticals - Government/Military, Telecommunications, Aerospace, Petroleum, Financial Services, and Academia.
Bill quickly went through the detailed responses from each vertical, then drew out common themes.
On the Business side, common themes were establishing the business case, the need for standards of Workflow in the extended enterprise, organizational change, risk management, and viability.
On the Technical side, interoperability, security/privacy, manageability, and network capacity and QoS issues were common concerns.
Bill noted the tangled Web that Grid appears to represent, until we virtualize it. The enterprise is becoming increasingly virtual. Agency Theory defines the enterprise in terms of its core value proposition and its relationships. Service-oriented architectures have the potential to greatly lower the barriers to establishing inter-enterprise relationships.
Critical success factors are seen as fluidity to be responsive to organizational structures and roles, flatness for flow of information , trust in collaborative activities and in reputation/risk matters, and culture to be willing to work together to accommodate and resolve the differing issues that arise in each organization.
How can The Open Group participate? Bill considered we in the Customer Council and Forums are uniquely positioned to address issues of architectural coherence, customer requirements, possible sponsoring of technology development, conformance testing and certification, and we can also encourage open source and Grid technologies. Although the technical issues are daunting, the real challenges of the Virtual Enterprise relate to human and organizational factors and we can all relate to resolving these.
Allen suggested other members may like to input and engage with the Customer Council to get involved with this - if so they should contact either Bill or Carl Bunje.
In answer to a couple of questions, Bill advised that we had not obtained any response to our survey from a Healthcare provider, and the lack of a common definition for what a Grid is was borne out by the survey responses in that perceptions are quite different depending on where you come from. The analogy with an elephant, made in a presentation on Monday remains apt - Grid is huge and it depends on your particular view ... rather like anyone touching an elephant, the view is different depending on which part you touch.
The Grid Reality |
|
Bill
Claybrook
|
Bill began his presentation by explaining that the Aberdeen Group is a small marketing consultancy, now looking at Grids from a user point of view.
Bill is interested in what is available today and what it can do, rather than what the larger potential is. The current business case is in the middleware area, all basically proprietary stuff.
Grid computing today is much more in the scientific/technical world rather that commercial world. Commercial uses tend to be limited to data-intensive problems and some transaction-oriented workloads. The question here is whether changing to Grids for these computing workloads is worth the cultural change and investment costs. Is the technology sufficiently mature? Oracle and Avaki suggest yes, but Operating System vendors suggest otherwise.
Bill asserted that enterprises are using server consolidation (moving many small servers to one or more larger, more powerful servers) as a means to reduce total cost of ownership). This means moving workloads to Linux on mainframes, Linux on blade servers, etc., using new hardware and new operating systems in many cases. Recent research indicates 1 in 4 servers is devoted to running production workloads so there is much real spare capacity available for Grid-type use. The Grid seems well-matched in both business and spare capacity in existing computer resources for supporting this move towards consolidation.
Considering the issues surrounding commercial Grids, Bill brought out that business will not move to use Grid unless and until clear business advantage is demonstrated. This has not yet been done but must be done before there will be significant changes in attitudes towards adoption. Grids are interesting in HPC applications such as research and development, and in high-intensity computational functions, but they have not been demonstrated as providing any advantages in commercial applications. Bill recalled Digital's millicent project for small lightweight transactions - would this be a good application to run on a Grid system? What kinds of applications make sense to run on a Grid?
Bill moved on to consider what Grid-enabled means, and how you can place applications in a Grid-server environment. All these questions need answering, and the simpler the answer then the more likely will Grid appeal as a technology that business customers will move up to.
Q: Eliot Solomon - He feels there is confusion over whether Grid is one virtual computer that does everything - e.g., transactions and/or batch work, or clusters of computing resources that do certain types of computing work well, and maybe a discovery scheme that finds the right resources, so how does the speaker view this confusion?
A: Gregg Ketmann (IBM) replied yes - some people see Grid as computer resource scavenging.
A: Bill - It is not possible to answer this question in a meaningful way because today the Grid is one virtual computer, but the necessary management and security software infrastructure are not yet available, the tools to Grid-enable applications are not yet available, and no one view of the Grid model fits all required perceptions.
A: Eliot responded with his view that the proponents of Grid need to explain the application areas that Grid is best suited to and demonstrate how well and how much more efficiently it does it.
Information Technology: Avoiding Liability and Increasing Value |
|
Michael
E. Bieniek
|
Michael is in Mater, Brown, Rowe & Maw's IT Practice Group, which does not do litigation. He noted Ian Foster's keynote presentation in Monday's Plenary when Ian said that IP issues have the potential to spoil everything in this Grid movement. Copyright and patent law has roots in ancient law. All Grid-environment applications are not explicitly covered by existing law. Licensing properties and agreements are also affected by Grid, but again there is no explicit coverage in existing law. The same is true for Internet and Web publication. This is in fact generally the case in the land of software.
In these situations, the initial response by lawyers is to attempt to shoehorn the new environment into existing law. What ultimately happens is that the law-makers subsequently get round to revising the law to accommodate the new technology. We can expect that to happen for Grid. Yesterday's speakers anticipated 1 to 4 years for take-off of the technology - by comparison we can expect the law-makers to lag up to 40 years behind.
Michael considered Intellectual Property Rights (IPR) in software and data:
A well-structured licensing agreement addresses the issues of:
Licensing saves you from penalty provided you ask the "who what why when where" (leave the "how" to the technical people) questions and provide acceptable answers. Licenses 10 years ago did not include outsourcing and Web services enablement because it was not thought of 10 years ago. So if something new is required to be done with the software that is not covered in the existing license, then that license needs to be updated to include that new something.
The same applies to Grid-enablement. The "where" is going to be an interesting emerging issue here - in Grid there is no "where".
Michael also recalled that in Monday's Plenary Wolfgang Gentzsch said you get what you want, etc. - SLAs will be very complex here, because Wolfgang also said you don't care where it came from. However, the law says "oh yes you do care where it came from".
Q: Mike Jerbic - What is "fair use" under the law?
A: MB - This is use that does not infringe what the law was intended to prevent.
Configuration Management - A Foundation for Grid Computing |
|
Steve
DuScheid
|
Steve explained that Tally sees five layers of functionality for a Grid to operate:
In this presentation Steve said he will focus on the Infrastructure Resources layer.
Trends are towards the extended enterprise - extranets, contractors, web services, outsourcing; and occasionally connected computers - mobile workforce, wireless connectivity, handheld devices. The extended enterprise encompasses many variations and new ones will continue to show up as long as people dream up new ways to use technology.
Management includes availability and performance as well as configuration management. In configuration management we have installation, upgrade, support, discovery, and tracking. We need to keep checking what you're keeping is right, and the principles you are using to check you are achieving the goals. These goals are to keep users productive, keep software operating, and keep hardware running.
Steve continued with a review of the dimensions involved in configuration management, covering hardware, software, and users, and listed the elements involved, including overlaps in each of the three categories.
A key part of configuration management is auto-discovery. This function needs to have a complete view of the infrastructure, and discovery mechanisms to identify resources on each machine. Some standards help here, to update the knowledge base the configuration management (CM) system maintains. New resources are not usually self-reporting when they are added to a system, so the CM system needs to do its own discovery and updates. Distribution of applications, upgrades, patches, registry change, and new virus definitions are also a part of the CM role.
Self-healing and state management also needs to be maintained, to ensure the desired state of each system. Remote control is the key to reaching out to customers for efficient and timely support actions. Software use analysis is important to ensuring you run efficient cost-effective license management, checking entitlements, and paying only for software your business really does use.
In summary, CM is highly relevant to the new emerging Grid environment.
Solid Policies and Procedures - Essential for a Successful Enterprise Grid |
|
Donna
Johnson Edwards
|
Donna said Tenax consulting is in the business of helping business enterprises to recognize IT legal and asset management needs in their systems. Solid policies and procedures are essential in an enterprise Grid.
IT asset optimization includes legal compliance (licensing and copyright, use of Internet and email, and data protection and privacy), and cost control (maintaining the value of hardware, licensing of software, and generally the total cost of ownership). Legislation is extensive, and legal penalties for non-compliance can be severe. Emerging laws are carrying even more severe penalties.
What's going on in your organization? How do you match up against the law? Piracy statistics are alarming. Do you have compliance controls, and if so how do you know if your controls are effective? You need to think in terms of providing evidence that stands up in a court of law. Internet compliance is also a big issue - could you defend your policies, procedures, and procedures? Likewise e-mail compliance. Regulatory compliance is also a major issue.
PDAs are another risk; data protection and privacy yet another. All these need to be covered by appropriate policies and procedures that are enforced by an effective management. Who is responsible in your organization for violations and penalties? These issues are worth checking on.
Formalizing compliance process is an important issue. The goals of compliance are to create a sustainable economic and legal IT environment. There are key stages involved in establishing this. Donna listed the kinds of policies and procedures involved. She went on to consider auditing, reconciliation, and ongoing management and updating.
Why is all this not a top issue? It needs to be. The Tenax model takes in legal, IT, and business issues in its whole approach.
Asset control is Donna's third big question. TCO and reducing cost of ownership is a key business driver. This also maps onto the Tenax solutions model.
Q: Chris Greenslade - Grid computing is not limited by state laws - what's the impact going to be at international level?
A: DJE - There are treaties between jurisdictions, and also there is more push towards standards which are often accepted as having international standing. Maybe The Open Group should set up a Forum to address this? The European Union respects the US Safe Harbor program which provides some protection.
Globus' Role in Development of Grid |
|
Dr
Jennifer M. Schopf
|
Jennifer outlined what Globus is doing in development of Grid. It sees itself as problem-solving in the 21st century - taking sharing to the next level. Existing technologies are helpful but not complete solutions. What's missing is support for sharing and integration of resources, in dynamic, scalable, and multi-organizational settings.
The Grid is the solution. Building the Grid requires four main components: open source software, open standards, open communities, and an open infrastructure. Globus is at the heart of this - creating an infrastructure and the mechanisms needed.
Jennifer went on to talk about how Globus sees and runs their projects. All their activities contribute to their common mission: in research, software development, application consulting, and infrastructure consulting. Globus is really a huge community, built on open source and openness. She described their approach to driving progress, and reviewed their growth on the early Grid history - dating from 1997 to the end of 2002.
The consumers of Globus' Grid work cover a huge range, and a similarly large range of application areas. Many adopters of Grid solutions need such very high levels of computing power that they have no option but to use Grid to accomplish their task.
Jennifer traced the adoption of early standards that have formed the basis for Grid, and how these have evolved to produce OGSA and OGSI.
She then gave a brief overview of what is in the Globus Toolkit GT3. The list of new services is growing rapidly as contributors suggest more features. An example is a reliable file transfer service.
She concluded by indicating the next steps and goals for Globus through 2004.
The GRIDS Center - Helping to build the infrastructure |
|
Tom
Garritano
|
Tom described what the GRIDS Center is and does to help build the Grid infrastructure.
National Science Foundation (NSF) Middleware Initiative (NMI) involves two major teams, and GRIDS (Grid Research Integration Development and Support) is part of NMI. The participants are essentially academic institutions. Essentially, NMI strives for an architecture and approach to middleware that can be extended to Internet users around the world. Their outreach includes strong links with other initiatives that are building their Grid software.
Tom reviewed the coverage of the GRIDS Center software suite. Major NMI software releases are currently scheduled in April and October each year. He then covered some Grid history which Ian Foster had covered - pre-Internet, etc., in the context of leading into their emerging cyberinfrastructure. He went on to identify several model Grid deployments around the world.
Regarding future GRIDS plans - Tom explained that their original three-year warrant takes them through to Fall 2004. They are hoping their excellent results to date will win them a favorable response to their request for an extension, so they can continue their work in support of Grid. Tom also mentioned GlobusWorld-2004 to be held in San Francisco, Jan 20-23, 2004 - see www.globusworld.org.
Global Grid Forum |
|
Dr
Jennifer M. Schopf
|
Jennifer presented a GGF primer explaining in her slides the GGF objectives, mission, their working groups and research groups, and the areas they are currently active in. She described the GGF document series modeled on the IETF RFC series. She then moved on to describe the GGF governance - its inception and evolution, its chair and steering group, its members and participation, and its advisory committee. Jennifer also introduced the role of the Grid Research Oversight Committee (GROC).
Participation on the GGF has steadily grown, and for the last two conferences attendance has reached about 700. They meet three times each year, and participation by anyone interested in actively contributing work is welcome. Their next meeting is GGF-9 in Chicago on 5-8 Oct 2003.
Boundaryless Information Flow and the Grid |
|
Dave
Lounsbury, Terry
Blevins, Eliot
Solomon
|
Terry Blevins (CIO, The Open Group) concluded two key things from this Grid Plenary meeting:
He said he has heard from a Grid expert that Grid sounds like boundaryless computing rather than boundaryless services. It is dealing with many issues that Boundaryless Information Flow represents. Our challenge is to identify and leverage the complimentary parts of both, and set up a cross-organizational project to make this happen.
Dave Lounsbury (VP Research, The Open Group) appreciated the scale and scope of Grid, and its consistent theme - use of open standards and open software to achieve its goals. As Grids gather acceptance and work together, they will pose real challenges which are truly interoperable. The strengths of The Open group are in open standards and in certifying conformance with standards. He put forward two challenges:
Eliot Solomon noted with some pleasure the complimentary mix of vendors and users in this Plenary. The ultimate goal of Boundaryless Information Flow is to improve business. However, we need to be more specific in defining how. He liked Greg Astfalk's and Bill Estrem's likening Grid to an elephant - to a blind man, feeling different parts of the elephant gives him a different understanding on what it is, but no understanding is wrong.
The same applies to Boundaryless Information Flow, which is about making IT business work better but within reasonable cost and burden. The challenge here is that this goal is very broad, and many architectures already address parts of it. Our goal is to develop reference architectures that bring the broad principles of Boundaryless Information Flow to bear on tangible business goals, and that give usable support and guidance to business decision-makers as well as to IT architects and technologists.
To achieve this goal we have created six models of Boundaryless Information Flow that address specific business goals, and we are asking people to respond with their IT approaches to each model. From their responses we will construct a reference architecture for each approach.
Eliot proceeded to introduce and outline the particular features of each example, each of which has different business requirements for boundarylessness:
He then suggested some common systems architectures that may be relevant:
What we're trying to accomplish in this work is to help software architects to meet their challenges, and help buyers make informed choices. To make it real - our goal for the October Conference is to identify specific technologies, products, standards, and architectures that address the specific models of boundaryless; hold workshops/presentations at the Washington Meeting at which the salient features that create “boundarylessness” are identified, compared, and contrasted; and make plans for next steps; e.g., standards development, conformance testing, creation of guides, etc.
We want all attendees here in this presentation to participate. To get involved, contact us at boundaryless@opengroup.org.
Home · Contacts · Legal · Copyright · Members · News |
|||
© The Open Group 1995-2012 Updated on Wednesday, 6 August 2003 |
|||
|
|