| |
Wednesday 6th August 2003
COMMENTARY - WSDM THE MISSING LINK
| LINKS
CBDI Newswire Tuesday 22ND July 2003
HP FUELS INDUSTRY-STANDARD APPROACH TO WEB SERVICES MANAGEMENT
HP has announced its plans to submit its Web Services Management Framework for industry review to the Organization for the Advancement of Structured Information Standards (OASIS), a global consortium driving the development, convergence and adoption of e-business standards.
CBDi Report
CBDI Newswire Thursday 6th March 2003
COMMENTARY - WEB SERVICE MANAGEMENT MARKET MOVES
Abstract: The Web Services Management (WSM) platforms and tools market is a wreck waiting to happen. Too many vendors with early stage products, no standards and not enough customer demand yet. This is all about to change.
CBDi Report
CBDI Report January 2003
ROADMAP REPORT -
REENGINEERING SERVICE LEVEL MANAGEMENT
Web Service management platforms enable monitoring and manipulation of Services at execution time and enhanced service level management. Clearly this is essential to manage the Service Level Agreements (SLA's) covering federated Services, in order to deliver, and to be seen to deliver trusted Services. However managing adherence to SLA's is only the start of what's possible using active run time Service management, and the same technical capabilities can also be used to deliver enhanced business models. In this report we discuss the requirements and provide a framework for using the new Service Level Management (SLM) technologies. Report available to Silver and Gold members.
CBDi Report | ABSTRACT: Management is such an abused and misunderstood term. Yet Web Services will not be used pervasively until we have platform neutral management. We assess two recent and very important announcements in this area, and revise our forecasts. Note: For background to this important subject, see links below.
It has been clear for some time that the primary inhibitor to widespread adoption of Web Services is lack of management capabilities. A Web Service is only going to be used if it can be trusted, available when needed, reliable in execution, meeting agreed service level commitments.
Although there are many vendors offering Web Services management capabilities, few users are committing to them yet. Most Web Service users are implementing short term holding actions pending entry of major vendors, the inevitable shakeout in the management products market and standardization of the management protocols.
To date we have been relatively cautious in our advice in this area, because progress has been very slow both from the major vendors and in the standards efforts. However things are now starting to happen. Last week we reported on HP who published its Web Services Management Framework and announced plans to submit same to the OASIS WSDM Technical Committee. (See Link below) This week we report on a parallel initiative by IBM, CA and Talking Blocks. We have talked to all of the submitting parties and are cautiously optimistic that this area is now about to move into a higher gear.
IBM, CA AND TALKING BLOCKS
IBM, CA and Talking Blocks (ICT for short) this week briefed us individually on their intentions to submit a specification to the OASIS WSDM TC. The specification is not yet available, but the essential elements are as follows:
- The ICT specification follows the principles established by other WSxx protocols. It is extensible, defining the service interface only and is entirely implementation and infrastructure independent.
- The ICT model is based on expanding the existing interface, where the management properties become an intrinsic part of the Web or Grid Service, as opposed to being a separate agent or object.
- The ICT model is also open. It does not imply a specific management framework and allows the management protocol to be rendered in various standards including WSDL 1.1, GWSDL, CIM and WSDL 1.2. (NB CIM is the Common Information Model from the DMTF and GWSDL is Grid WSDL from OGSI)
A member of the ICT group indicated they have looked carefully at the impact on the developer, and have implemented the specification as a proof of concept on .NET, Axis and IBM environments.
HOW DOES THIS COMPARE TO HP'S SPECIFICATION?
The genesis of both the HP and ICT efforts was actually in the W3C Web Services Architecture Group where the parties to both submissions collaborated on a management taskforce defining a set of requirements. Since then the parties have gone their separate ways. HP's approach is somewhat different insofar as they have defined a complete framework and infrastructure. Two areas of the model stand out as being potentially quite different to the ICT model; a) the HP model is based on a managing agent approach, which may imply a different set of operations and b) the HP model has implemented specific event support. However all the parties are very optimistic that solutions to issues will be found and resolved.
WHAT DOES THIS IMPLY FOR THE WSDM PROCESS?
Some commentators will always try to find controversy. They would like nothing better than the HP vs ICT situation suggesting that IBM is exerting undue pressure by submitting a second specification and creating competition in the standards development process.
Actually this picture is completely incorrect. There is a little competitive jostling going on, but it not as might be expected between IBM and HP. For the most part the participants are taking an altruistic position which recognizes the overwhelming importance of achieving a high quality version 1.0 standard as more important than promotion of specific IP. Our assessment is that consensus will be achieved reasonably swiftly. For example last week at the OASIS meeting HP committed to OGSI support.
Our reading of the situation is that a) the specifications have been submitted to the same standards organization, b) there is an obvious intent from all parties to achieve a standard as soon as possible.
The target for completion of the WSDM specification is January 2004. On the basis of our conversations with all submitting parties, we assess there is high probability this will be met.
WHAT ABOUT MICROSOFT
Notable by their absence from this important TC is Microsoft. Evidently they were invited to join the committee, but indicated they were "not interested in management at this time". We might interpret this response as MSFT is not a major player in this area currently; further they do not see management as a contentious area and are happy to leave the development to others. In addition they might actually be pleased to be able to point to areas where the IBM/MSFT hegemony is not dominating the standards development process. We do expect MSFT to get involved in management through at least the WS-I process.
PLANNING FOR WSDM IMPLEMENTATION
The standards that emerge from the WSDM committee will certainly have some significant impact on architectural thinking. First there is the open question about the managing agent or native Web Service models. We heard two opinions here, a) that current protocols enable multi-hop architecture, and that there is no further work to do, and b) that there will be a requirement for additional operations. It's a reasonable assumption that the standard will support both, but for users there is an important choice of architecture to be made, which may have major impact on performance.
Second we have the whole question of performance. Regardless of architecture there's going to be a huge amount of extra traffic generated by management operations. So if you are already worried about the performance of (unmanaged) Web Services, it's time to get "really" worried.
Performance is also the area where vendors can do their level best to differentiate from their competition, and how the individual vendors respond to this challenge is going to be a subject of great interest over the coming months. This is going to be the real competitive arena. Speaking to the various vendors already we see differing solutions emerging, reflecting their increasing focus and concern on this topic. The range of options under consideration includes:
- implementing the management interface in the Web Service, but probably only appropriate for smaller, less performance critical applications. This will drive demand for XML accelerators etc
- aggregation of data - collecting together sets of management operations to minimize traffic
- implement the management interface in the application or a specific service container
- providing native proxies for specific platforms
- reverting to non XML protocols under the covers
All of the submitting parties fully expect considerable physicalization. It must be said that all parties expect these issues to be resolved, but our assessment is that performance will be an issue through at least 2004.
Another interesting idea from CA is that they are actively planning hosted management of Web Services. They anticipate this will be a strategic market opportunity for them to assist enterprises to avoid infrastructure investment costs and to move forward rapidly with trusted Web Services, and are planning hosted services in the relatively near term. CA also confirmed they would be eating their own dog food and plan to provide SLA's on these hosted services using their own hosted services.
CONCLUSIONS
So when will WS management get real? In our guidance to date we have been fairly cautious about management, because there has been no real visible activity beyond the formation of the WSDM committee. Our working assumption in our protocol stack timeline report has been to plan for early adopting activity in 2004. With this weeks news we can be more positive. The deadline for the TC is January 2004 and there seems to be cautious optimism that this will be met with a minimum necessary specification. Consequently it is now realistic to plan for early adopting activity in the 1H 2004. However we would caution that a) performance is going to be a significant area of ongoing development and this will need careful monitoring as we move forward and b) there will probably be ongoing protocol development. What functionality is not covered in the version 1.0 is as yet unclear. |
| | Read Full Article / Link to referenced materials |
|
Home