Home   

 
Home Page Editorial
(April 9, 2001)

From Publisher and Managing Editor's Desk...

Do initiatives like BREW, J2ME and MicroChaiVM Signal Something About Software Development Environment For Wireless Applications?

Qualcomm announced BREW - Binary Runtime Environment for Wireless application development, recently. BREW is a special SDK for CDMA networks. On a different front, Sun's J2ME is making significant inroads as a multi-platform development environment. Yet at another level, HP announced MicroChaiVM for mobile applications. Should network infrastructure providers create proprietary software development environments for their unique wireless networks or should we as an industry try to adopt a scalable multi-platform tool set? Here is MobileInfo's analysis of this important issue:

1. First of all, it appears to us that wireless and mobile computing industry continues to be fragmented in terms of standards in a number of ways. A few attempts are being made here and there but these are not commensurate with the enormity of the task and stakes. We do recognize that in a fast-paced industry like ours, it is difficult to wait for dejure standards before developing products and services. Defacto standards are more likely. WAP is one of the protocols and not the only one.

2. One of the most important areas where there is a definite need for standards is in the area of wireless network protocols and a software development standards because virtually hundreds of thousands of software developers are involved in this effort. 

3. Software development must be viewed at three distinct levels - software drivers, lower level horizontal applications (e.g. messaging) and then high-level business applications custom developed for the enterprise. Quite often, these applications assume the form of a general-purpose package that can be customized to a unique business process or processing logic. While there are common threads in all three layers, single set of guidelines do not apply in all cases.

4. It is becoming increasingly obvious to serious wireless and mobile computing systems integrators and technology architects that Utopia of a single wireless network is not going to happen in the near-term future. In fact, realistically speaking, it will be at least 7-8 years before a single network interface is likely. There will be GPRS, EDGE, TDMA, CDMA, W-CDMA and UTMS. Should each network supplier provide a unique and proprietary SDK for its own network or should we try to go along the J2ME route that Motorola has chartered.. We do not see any technical or business advantage to any wireless network supplier in creating a unique SDK for its network, howsoever unique it may be. Going above the chip-level software development, we feel that it is possible to create an SDK that utilizes a multi-platform environment like Java. In this context, we applaud Motorola in taking the lead in going with J2ME with its i-DEN network application development environment.

5. Once we have Java as underlying software development environment, network suppliers can provide software drivers for their unique networks so that vendors like Broadbeam can  develop communications middleware that support multiple networks. This means that corporate application developers do not have to worry about lower network layers. 

6. There are two parties to a communications session - the client and the application server. While the above may serve the server end, there is a similar problem at the device end as well. Multiple operating systems (Palm OS, EPOC, Pocket PC, specialized Embedded OSes for smart phones) and a variant of processors make the task difficult for application developers. Single universal microbrowser has not emerged. Should we leave it to the creativity of device operating system developers to give us a unique browser interface so that we can program applications to this interface? This is a reasonable compromise.

7. The task of m-commerce application developers is not getting easier. Different application servers with unique APIs, security software products with varying levels of application server support and not-fully-developed payment systems with unknown APIs pose significant challenges and learning curves. Experience shows that it takes at least three versions across 24 months before a software product reaches acceptable reliability standard. We must prepare the carriers, portals, ISPs and Wireless ASPs to understand this reality and prepare their technal support teams to roll up their sleeves and solve these problems before the original software vendor (akin to OEM) gives a reliable and functional product.

8. As far as enterprise business application developers are concerned, they are better off leaving this task to experienced application software development houses or rely on application packages that some vendor in its own technical wisdom has developed and made functional after trial and error. Look for a a working packaged solution and not create one from scratch with the best components but where inter-component compatibility may be a problem. While you can try buying a best-of-breed solution based on the best device with most functional OS, fastest wireless network, best application server, most thoroughly-tested middleware interfacing to your backend database and application system, you may have to compromise and accept tradeoffs in this process. In this environment too, J2EE might offer a preferred environment.

9. Finally, before you accept a unique proprietary environment, consider your corporate IT application development standards. If you are in an IBM environment, investigate seriously if IBM's WebSphere will do the trick for you.  If you have used Sybase's Visual development environment for your corporate application development, go with iAnywhere's development tools. Same is true with Oracle shops with OracleMobile. HP's e-bazaar may be attractive to HP shops. Stay within familiar environment after verifying that vendor has delivered on a blueprint that they promise in their corporate presentations.

10. Remember that wireless applications should be logical extensions of your wireline and Internet applications. By this we mean, information roots should lie there but look, feel and user interface will have to be uniquely modified to meet the limitations of networks, devices and mobile business processes. You should have a holistic view and not an isolated view of your applications.

Chander Dhawan - Your Site's Principal Consultant and Publisher

Do you want to make any comments on this editorial? Send us an e-mail.