Fixed connection networks, by design, provide an effective platform for file and resource sharing or group interaction within (relatively limited) local areas. The architecture of these networks is optimized for this environment-assuming the presence of a dedicated, high-bandwidth, highly reliable network connection, which incurs no incremental usage charge. When applications developed for these networks are deployed to large numbers of users who are not continuously connected by a high-bandwidth, highly reliable network connection, they exhibit inherent limitations that become increasingly evident as the number of users increases.
Challenges of the Mobile Environment
Those responsible for supporting mobile users within the "extended enterprise" must become acquainted with a handful of key factors which differentiate the mobile working environment.
Fixed connection LAN-based computer users typically have the use of feature-rich multimedia devices which benefit from substantial processing power, and unlimited, continuous access to information, applications, and systems and are not dependent upon battery power. They rarely have to worry about synchronizing client and server data, are locally managed and supported, and have come to expect fast, cheap, and reliable network connections.
Compare their lot to that of the mobile worker; they typically have light, small form factor, low-cost devices with limited memory, processor capacity, and battery life. These users have to deal with constrained, intermittent access to corporate systems, suffer unpredictable connection windows, and as a result need to do much of their work offline. They have to worry about how and when client and server data will be synchronized, generate complex management and support problems, and labor under slow, unreliable, and expensive connections.
Using GSM, the most widely deployed mobile data communication technology with in excess of 300 million subscribers, as an example, the maximum possible connection speed of 28.8Kbs is more than 300 times slower than the minimum of 10Mbs that is normal for a fixed connection local area network. Add to that the thought that mobile users simply do not want (and cannot afford with current billing tariffs) to be online all the time and the throughput gap widens to as much as 1:10,000.
Despite the promise of 3G telecommunications standards, and broadband wireless technologies, no one should imagine that these order-of-magnitude bandwidth limitations are going to disappear any time soon, or that the provision of faster and more reliable wireless communications will be any more cost-effective than today’s solutions.
According to Gartner Group and others, the majority of mobile users will still be connected to the enterprise at speeds that are hundreds of times slower than a LAN connection until at least 2005. They will have to remain disconnected most of the time. They will not have the luxury of calling on IS help from "down the corridor" if they have a problem. Finally, unlike the fixed connection LAN where the bandwidth is effectively "free" once the system has been installed, mobile users will be paying for every second of connect time, or packet of data they use.
Challenges Facing the Enterprise
It is clear that the management and support of mobile workers requires a new architectural approach, an approach that focuses as much on keeping them productive while they are offline as it does on providing an efficient connection when they are online. Systems administrators want to be able to manage mobile systems and applications effectively, reduce cost-of-ownership, and maximize end-user productivity.
Unfortunately, while most organizations have improved their computing resources and capabilities at their centralized sites via a mixture of client/server and intranet technologies, few have successfully implemented these systems in conjunction with the vital information flow of their mobile workers. Systems that are implemented for use by mobile workers while clinging to LAN-based assumptions have been proven to produce poor results. Applications that are straightforward to deploy in a LAN environment become more difficult to prototype, build, and deliver to large numbers of mobile users, unless the application development tools provide the applications services that these environments require. As a result, cumbersome paper-based processes are too often used to address the information management requirements of the mobile workers, to avoid the cost and complexity of developing and delivering systems designed and optimized for this unique environment.
Today's applications, built to work on PCs over a high-speed communications infrastructure, do not in general work well when applied to mobile devices over the fragile communications infrastructure which characterizes the wireless world. These demanding implementation, usability, and manageability issues threaten to constrain the growth of the mobile data and mobile commerce sector. Some innovative proprietary solutions exist, and there has recently been much effort from several industry initiatives seeking to standardize an architecture which incorporates the needs of mobile wireless devices, but too often these efforts have been focused on delivering point solutions rather than considering the global picture.
The industry has been paying attention to the increased burden of Total Cost of Ownership (TOC) for bandwidth-challenged mobile and wireless devices, Gartner Group believe that the percentage of IT budgets allocated to supporting these users will grow from less than 10% in 1998 to over 30% in the current year.
This increased cost is due to accommodating a set of requirements from enterprise wireless and mobile end-users, an insight into which was gained during The Open Group’s research and interviews and the following is a sample of some common themes:
- Maximizing customer face time
- Minimizing administration and other distractions
- Having critical information to-hand and up-to-date
- Having the ability to recover critical data when a system failure or theft occurs
- Being able to connect easily from whenever and from wherever I like
- Remote support-not having to “return to base” to resolve problems
- Minimizing time wasted waiting for connections to complete
From an enterprise management point of view, the most commonly stated requirements are similar, including:
- Timely deployment of applications and their updates to wireless users
- Ensuring critical corporate information has been distributed to the appropriate users
- Having access to an accurate inventory of all hardware and software deployed
- Ensuring effective backup for all critical corporate data on wireless devices
- Providing cost-effective, remote support
- Making the most effective use of the communications budget
All of this has to be provided in the context of the more challenging mobile and wireless environment.
As a result, the architecture of fixed connection network systems does not provide a complete range of capabilities required for effective enterprise-scale wireless and mobile computing solutions. They lack the economy of operation, transparency of underlying management during a connection employing unused bandwidth, ease-of-use, and administrative flexibility required to automate core commercial business processes successfully for large numbers of wireless and mobile users.
One characteristic that all network connections have is the need for session management. That need is even more evident and pressing in an occasionally connected model. This needs to be independent of the underlying transport layer (but needs to be cognizant of it and exploit such benefits as the respective transports provide).
The importance of session management to mobile architectures was recognized by most of the individuals and companies interviewed. The feedback received during the research points to session management being a key enabler for wireless data interoperability in the enterprise environment. Considerable benefits can accrue-both to the vendor and user community-if a standard approach is established for this highly varied and complex foundation of all mobile and wireless network processes.
This report posits that the functional requirements (often integrated within the logic of a particular application) of a layer to manage the session, could be abstracted to form a general, open architectural layer to support the wide range of other management services such as data synchronization, file and software distribution, accounting, security, content distribution/transformation, and so on. It is critical that the standard implemented ensures interoperability by having open external interfaces for the support of multiple applications and is agreed by all the major players in the wireless and mobile market.
The opportunity exists for The Open Group Mobile Management Forum (MMF) to accept the formidable challenge to define and gain acceptance for a common session management architecture. Why formidable? Because the MMF will have to persuade vendors to help define and agree to a standard reference model offering a protocol to support services which specifically address the unique requirements of the mobile user, as well as interoperating with existing initiatives such as Distributed Management Task Force (DMTF) and Intel’s WfM (Wired for Management).
A basis for discussion by members of the MMF working on a common session management reference model could include the following characteristics.
Integrated Approach
An integrated approach is required for a complete wireless and mobile computing solution which not only spans enterprises, but allows information exchange between all members of the electronic trading community. A complete architecture must be independent of the transport layer and accommodate common transport protocols. It should offer consistent reporting of failures and allow for automated restart of a session in case of any failure. Remote control of wireless devices without a direct connection is clearly also required for remote device support. The protocol should provide a set of implementation rules that support and bridge the range and levels of application development protocols employed to build a solution.
Scalability and Control
As the market, and therefore the potential population, for wireless and mobile systems grows, the architecture must permit:
- Ease of expansion of system capacity (rapidly and economically) without forcing lock-in to a particular implementation
- Creation and deployment of applications rapidly as business requirements dictate
- Managing deployment of updates and revisions to existing applications
- Control functions to monitor and control information flow to and from disparate system users
- Minimizing operational costs by minimizing the amount of data transmitted and minimizing connection times
Ease-of-Use
The elements required to enable applications for wireless and mobile deployment should not require familiarity with particular operating characteristics and complexities of each device. Furthermore, they should shield consumers from complexities of data communications and be logically consistent with the consumers expectations of service. For example, functions that offer significant end-user transparency include sophisticated scheduling facilities (both at the server and at the wireless and mobile device) as well as a scripting facility to allow a variety of defined sequence of events with in-built logic to be defined for different sessions. To achieve transparency and shield the end-user from technical errors and communications problems, the standard should provide automated condition handling and reporting processes.
Compatibility with Existing Systems
The protocol should also offer open integration for most existing computing and communication environments in order to leverage existing customer and third-party investments in these technologies, while providing interoperability between various disparate members of a trading community. Examples are:
- Access to a range of widely-used database management systems
- Support for industry standard and product-specific application programming interfaces (such as the evolving SyncML initiative, for example)
- Adherence to Internet standards, such as mail, and so on
- Access to the variety of communications types necessitated by geographic distribution of wireless and mobile users
Communications Control
The standard reference model or framework should make it possible for solutions providers to automate and manage the exchange of information between applications and databases that reside on host systems with wireless or mobile users regardless of their location, or the wireless device type they are using.
It should provide methods for selecting, controlling, and minimizing communications, which includes:
- Central scheduling and control of communications sessions
- Provide underlying systems management transparently during a connection employing unused bandwidth
- Accommodate multiple communication alternatives
- Minimize cost and time
- Provide detailed audit trails of system communication events
(… from a historical platform perspective)
To date, the principal operating systems manufacturers have all failed to embrace the full set of requirements adequately, and implement a complete architecture that serves this category of users effectively, allows interoperability, whilst giving appropriate freedom for creativity and competitive advantage, and hence consumer choice. Many suppliers are recognizing that a three-tiered model providing a (store and forward) function point between client and server to accommodate the peculiarities of the remote and mobile world is required. (This has also been referred to as a "queued architecture".)
Leading platform vendors have developed several such architectures. For example:
- IBM’s Message Queue architecture comes close to achieving this aim but in the design has done nothing to improve the interoperability situation as it is proprietary and requires a significant amount of custom development.
- Microsoft came close to adopting a store and forward model in the mid nineties-until it was faced with the significant distraction of the Internet-although recently Microsoft has stated that they intend to put into the soon-to-be-announced version of CE (code-named Rapier) a version of Microsoft Message Queue Services (MSMQ) to better support distributed applications.
- As an example of the telecommunications industry wrestling with this problem, in the last couple of years Bell Communications Research has recognized aspects of the requirement above by reflecting the three-tiered architecture model in their approach to wireless mobile network computing.
The ability to deliver a wide range of critical information types selectively, transparently, and effectively to the appropriate people in a timely, controlled, and reliable fashion, wherever the recipients of such information are located, and whenever they choose to connect to the host system, is a necessary component of a successful wireless and mobile implementation.
The quandary facing most corporate IT organizations with the range of offerings available today (even if they are complementary functions) is which of these independent approaches should be used as the infrastructure choice for their organization. Differences in architecture, structure, and service between the various approaches available in the market often mean incompatibility and obstacles to interoperability. This forces complexity and resulting delay as regards strategic technology selection decisions, which in turn act as a brake and an inhibitor to market growth at this critical stage of market evolution and acceptance.
The lack of architectural fluency in session management is a contributing factor to the poor reputation, which some enterprise Mobile Computing projects have acquired in recent years. As the enterprise extends the reach of its applications-beyond the relatively "safe" environs of corporate users-to include business-to-business and eventually business-to-consumer offerings, the impact is likely to be even more severe.
The Open Group’s Mobile Management Forum (MMF) should utilize its position with suppliers and enterprise buyers to work with the various vendors of existing management products serving enterprise use of Wireless and Mobile Computing in order to agree to a Common Session Management Layer (CSML) which would certify interoperability and compatibility of architectures.
The proposed Common Session Management Layer (CSML) should be exactly that-common to all sessions between the wireless device and its communicating partner. Herein lies the daunting complexity, as any attempt to standardize this layer needs to strike a balance between the rigors required for a standard and the flexibility to accommodate the vast range of functions and services that would need to operate on top of this layer. Beyond fairly obvious examples, such as email sessions and management functions (for example, software distribution), a broad range of e-business functions (payment, receipts, reservation confirmation, and so on) would need to interoperate with this layer.
Adoption of CSML in future releases of wireless data products would simplify the implementation choices of corporate IS departments, thus accelerating the widespread adoption of Wireless and Mobile Computing solutions, and generating continued growth for the industry.