Advertisment

Will We All Speak E-speak?

author-image
DQI Bureau
New Update



What does it take for something new to catch on? Why is it common for seemingly good technologies to fail? One of the trends in application

development tools, computer-aided software engineering (CASE), hit the market in the mid-1980s, and for years was hyped as the thing that would reinvent software development. While many factors contributed to the overall tepid response to CASE and to some major failures when expectations swamped reality, the main problem was that CASE did not in fact

fundamentally change the way software was designed and developed. More accurately, CASE automated aspects in the use of inherently cumbersome systems analysis and design techniques. It allowed users to draw with the computer rather than with pencil and paper. CASE’s productivity and quality gains failed to meet expectations, the adoption cycle stalled and the market shrank. Eventually, many elements of CASE did reemerge in a new generation of visual development tools.

Advertisment



We could say that the majority of today’s ecommerce sites simply automate traditional models for transacting business, such as electronic versus paper catalogs, purchase orders shipped over the wire, or agents versus people making approvals. Nonetheless, the adoption of ecommerce has been rapid and, by most measures, highly successful. While the internet and ecommerce solutions haven not reinvented how business is conducted, they have slashed the cost of transactions, improved quality, reduced cycle times and improved organizational responsiveness to changing business conditions.



Business models with an entirely new approach?



Think then of the potential of a next-generation internet technology–one that spurs rapid adoption of entirely new business models that would fundamentally transform how consumers and businesses behave with each other. E-services are self-describing modular service offerings that web-based processes invoke to perform a function. 



The industry will pass from its present stage to the next. The status quo includes e-tailing, customer self-service and supply chain integration with established business partners. That is, adapting traditional business practices to exploit the internet. The next stage would fundamentally change the way you do business.

Advertisment



The differentiators are:



  • Dynamic discovery, in which you find services by capabilities, not by name.



  • Brokering, in which service agents can facilitate a truly efficient marketplace to provide the best offering for your needs.



  • Composition, in which service providers can dynamically compose services from lower-level service offerings to meet specific needs.



  • Mediation, in which service requestors and providers can automatically manage, adapt and evolve in a dynamic services ecosystem.



A dynamic e-services ecosystem could well set the stage for a new, self-adapting market that can optimize both the pull and the push of electronic service solutions.

Advertisment



E-services scenarios



To illustrate the potential for how e-services could enrich business and personal experiences across a broad spectrum, let’s look at a few hypothetical scenarios before and after adoption of e-services:



  • REQUIREMENTS-BASED PRINTER SELECTION: Before–a network-attached computer user must manually select a printer to print a document. If the device supports only black-and-white and the document requires color, the operation fails or the user gets a printout that doesn’t satisfy her needs. After–pressing the print icon causes an intelligent driver to interrogate the document. It determines color, page size and resolution requirements. It submits a request to the corporate print e-service, which identifies the closest capable device and initiates the print. If none are available, it solicits bids over the internet from vendors offering print e-services, such as Kinkos and PIP, and returns cost and delivery proposals for approval. Bids falling within preapproved parameters are kicked off immediately, and UPS is notified for a morning pickup.



  • CORPORATE PURCHASING: Before–when a buyer needs 1,000 widgets he may phone a favorite supplier. He may call several suppliers to get the best terms and conditions. Or, if the company has already implemented internet-enabled buying, a purchase order is fired off electronically. After–with e-services the buyer sends a reverse bid specifying all the requirements for the purchase. Automated responses flow in from any and all potential suppliers who have advertised directly or indirectly through a broker. The purchase is consummated automatically, with credit and reference checks supplied through complementary e-service providers. Fulfillment and payment transactions are kicked off as a consequence. Alternatively, the buyer can directly contact a potential supplier for a large or specialized order or where contractual terms must be negotiated in a traditional manner.



  • ORGANIC HOME THEATERS: Before–$50,000 to $100,000 of exotic sound, screen and satellite connection boxes buys you a home 



    theater that becomes obsolete before you have a chance to spend another $5,000 to get it installed and the remote control programmed. After–media software firms provide generic, programmable hardware like speakers, components and wall-hung, flat-screen displays for free. You plug it into the internet–which is probably wireless–select and download the configuration and gadgets that you want, then pay for it. All programming is pay-per-view and payments are automatically transacted. Special effects are downloaded and billed as required for specific viewings. You specify your preference–for Elton John or Johannes Brahms–and new offerings are served up automatically by
    e-tailers.



  • ORGANIC HOME THEATERS: Before–$50,000 to $100,000 of exotic sound, screen and satellite connection boxes buys you a home 



    theater that becomes obsolete before you have a chance to spend another $5,000 to get it installed and the remote control programmed. After–media software firms provide generic, programmable hardware like speakers, components and wall-hung, flat-screen displays for free. You plug it into the internet–which is probably wireless–select and download the configuration and gadgets that you want, then pay for it.



All programming is pay-per-view and payments are automatically transacted. Special effects are downloaded and billed as required for specific viewings. You specify your preference–for Elton John or Johannes Brahms–and new offerings are served up automatically by

e-tailers.

Advertisment



Unlike many highly promoted vendor initiatives in the years gone by, the e-services initiative has solid technology underpinnings. E-speak products and components, and the services based thereon will encompass low to high level functionality. The true potential is some time out in the internet years.



Nonetheless, we think it is worth looking at and understanding today.



  • What is this technology, and is it a base from which the industry can move into the future of internet? 



  • What is the service and solution architecture that e-speak supports and what opportunity and value does it provide? 



  • What is the strategy required for achieving broad market acceptance, without which the ultimate vision will be unattainable?

Advertisment



The e-speak object model



The major objects comprising the e-speak framework are building blocks for future e-services enabled business channels. They are:



RESOURCE: The information asset enabled as an e-service through e-speak is called a resource. It can be an internet device, mobile phone, mainframe hosted application function or commercial electronic service. A resource can be self-contained and be able to directly implement the entire e-service or can compose–statically or dynamically–composite e-services from others.



Existing applications or devices are enabled as e-services by wrapping them with the e-speak client libraries and exposing or virtualizing

their facilities through the e-services interface (ESI) API. Resource handlers are instantiated at runtime to manage the interaction with the resource. A remote handler concept provides requestor transparency between local and remote resources.

Advertisment



E-SPEAK SERVICE: The e-speak core services are basic facilities within the e-speak architecture and are used to manage and provide access to resources. They are exposed through APIs. They include naming, registration, security, advertising, discovery and event management. The current set of services, while rich in functionality, is fairly low-level. Higher-level, value-added components that will roll out over the next several years will simplify and promote the adoption of the e-speak framework.



VOCABULARIES AND CONTRACTS: Vocabularies and contracts provide the means for consumers and providers to intelligently negotiate and conduct business. The contract is, in effect, the protocol that the two parties will use to converse. A vocabulary is the grammar used by a service provider to describe its offering and what a requestor must understand and use to discover and avail itself of the e-service. E-speak supports a default grammar based on the OMG’s CORBA Trader Service specification. Also, appropriate XML document type definitions can be imported as grammars through the e-speak Vocabulary Builder. Intermediate e-service offerings can translate or mediate the use of different grammars and contracts by requestors and service providers.



LOGICAL MACHINE: The combination of an e-speak core server, a cache of resource or service information and some inter-machine protocol handlers is called a logical machine. A logical machine manages and enables access to resources that connect to and register with it. It implements all the base services exposed through the APIs. In production environments it will also include an RDBMS-based metadata repository where service registration, system state and management data are maintained. It can be distributed across multiple physical servers, and multiple logical machines can be hosted on a single server. Groupings of related e-services will likely be hosted by ISPs and ASPs and access offered through logical machines established as points-of-presence or gateways.

Advertisment



CLIENT: A client is a process that has connected to an e-speak logical machine to consume or provide an e-service. Clients can be single or multithreaded.



E-speak APIs and services



Two APIs provide, respectively, client access to core e-speak services and inter-logical machine communications. Both are implemented as discrete interfaces into a common object model.



ESI: This provides the facilities for clients, including service requestors and providers, to connect to and access e-speak logical machines. A single client library implements both requestor and provider APIs to support bimodal clients that are required for constructing composite services. Requestor services include, for example, connection to a logical machine and e-service searching or discovery. Once an e-service is selected a resource handler is returned to the client, and this provides access to the provider’s offerings through the messaging and network object model facilities. The key service for resources is registration, for recording the availability and capabilities of an e-service. 



E-SIP: The e-speak service interchange protocol (E-SIP) implements the interfaces for interaction between multiple e-speak logical machines. It supports the export and import of e-service registrations, for example, as well as the instantiation and communication with a remote resource handler when an e-service is provided remotely.

E-speak logical machines are usually configured administratively, so use of E-SIP is restricted to the customization and extension of logical machine services. 



A robust framework



The scope of e-speak is broad, and will get wider and deeper over the next several years. The base services being introduced with version 1.0 of e-speak are fairly low-level but establish a rich, adaptable foundation. The quality of the design can be tasted by looking at a few of the services–advertising and discovery, messaging, and event processing–in more detail.



ADVERTISING AND DISCOVERY: The keystone of the e-services concept is dynamic advertising and discovery. E-services are described by their attributes, not by name. An e-service registers with a host logical machine. It, or its host local machine, can advertise the service by exporting the service description to remote logical machines. A higher-level advertising service supports publishing the service description to, for example, a commercial advertising e-service for broadband promotion and discovery support. Advertising implies pushing the service description to other logical machines that may have clients in the market for the service. The advertising service can be configured online or offline. Offline configuration uses LDAP and is designed to be compatible with any LDAP V3 compliant directory service.



Services are essentially described in terms of a named vocabulary and a set of attributes and values. Complementary discovery services allow a client requestor to search for services through its host logical machine. E-speak supports a rich and flexible query syntax for resource discovery. If the requestor and service provider understand different vocabularies, then the logical machine or a third-party e-service can act as translator. Whether a service is local or remote is transparent to the requestor. A logical machine can also dynamically query for remote service providers requiring preregistration with the logical machine.



MESSAGING AND THE NOM: The complexity of the discourse required to consummate an e-services transaction can range from trivial, like sending an email solicitation to a single user, to complex, like navigating all the elements of a complex integrated supply chain purchase. E-speak supports three protocols for requestors and providers to converse: asynchronous messaging, synchronous messaging and a network object model (NOM). The network object model allows an e-service to be manipulated as an object, with programming

interfaces generated from IDL. The messaging and NOM interfaces can be utilized individually or in combination to support almost any requirement. E-speak does not yet support message queue persistence or transaction semantics.



EVENT PROCESSING: Similarly, e-speak provides a comprehensive event service. It is implemented as a publish-subscribe subsystem layered on the e-speak messaging service. The event service is utilized by e-speak components for coordinating activities and for systems management, but it is primarily intended for push-oriented e-services. The event service involves four entities–Publishers, which create events and their content; Listeners, which consume events; Distributors, which capture and distribute events; and Subscribers, which manage subscription lists. 



NAMING AND SECURITY: Widespread adoption of a universal e-services market will demand flexible, easy-to-use, yet bulletproof naming and security services. E-speak addresses the first with a federated naming model–each requestor, for example, uses its own names for resources, and e-speak manages the translation and adaptation to remote resource names. This eliminates naming conflicts and incompatibility in name syntax across heterogeneous environments. In any case, resources are virtualized and are typically discovered based on attributes rather than names. E-speak provides a solid foundation for rich authentication and authorization services. Its security is based on capability sets and simple public key infrastructure cryptography, as defined by the IETF standards body. Resources exchange attribute certificates for authorization and names, so both services and users are authenticated through certificates. Once authenticated, resource access is controlled through access controls defined as capabilities and granted by the resource administrator to classes of requestors. Capability expresses permission to perform operations or obtain services. Any practical level of granularity can be supported. The security model is comprehensive, supporting service-specific policies evaluated against live data, but the current APIs are low-level. We can expect security software vendors and service firms to actively enter the e-speak market opportunity with high-level model abstractions and runtime services.



Prospects for success and e-speak’s role



E-speak will be most attractive to new and existing firms providing internet-enabled business services. An increase in e-speak usage will make it attractive for existing providers to wrap their offerings and provide them as e-services. The e-speak service architecture readily supports that.



A dynamic e-services environment requires more than technology. It also requires changes in the business culture, primarily the ability to fully trust a dynamic, often anonymous business model in which potential high-dollar

business transactions are formed, consummated and dissolved in microseconds. Will a user be ready to forgo the established supply chain to save a percentage point on major, business-critical purchases? This will require time. Intermediate services–for rating, guaranteeing services, payments or service quality monitoring–will need to be in place first.



In the meantime, some IT organizations may want to consider e-speak as an integration framework for their intranets or extranets. The resource abstraction model and overall technology infrastructure provide a technology and vendor-neutral environment for integrating application, data and services. If in fact e-speak takes hold, companies who adopt it sooner rather than later will certainly have a leg up.



Another evolution in the adoption lifecycle for e-speak will likely include sourcing. Major companies and third parties can offer yellow pages services based on dynamic discovery rather than static catalogues or directories. 



Out in front



One significant inhibitor to success with e-speak is the long, in internet years, adoption cycle that it will experience. People didn’t buy cars en masse until there were roads and a network of gas stations. Air travel required airports.



The universal e-services market will require substantial infrastructure investment. But unlike the massive and lengthy investment in a worldwide satellite network required for Iridium’s mobile phone service, the investment in e-speak solutions can be incremental and will likely be broad-based. While it will achieve a revolutionary shift in how e-markets work, it can be adopted evolutionarily.



Sun’s Jini is similar in concept but is much smaller in scale. Microsoft’s BizTalk addresses many of the same collaboration and vocabulary issues but at present it does not bring the same solid distributed deployment environment to the table. Alternative solutions are bound to surface. The problem is too real and the opportunity too big. One nice thing is that, even if there are eventually two or three solutions of the same ilk, we should have transparent

interoperability among them. Otherwise, one or the other hasn’t lived up to the promise of the universal e-services platform.



Richard C Atwood



Courtesy: Hewlett-Packard

Advertisment