Opening a window to the world

Most organisations wishing to open up some of their legacy systems to the web do not wish to change their applications, writes...

Most organisations wishing to open up some of their legacy systems to the web do not wish to change their applications, writes Nick Enticknap, who explains the practical options available.

Talk e-business, and people think of exotic B2B markets and spectacular B2C applications. But for many companies, the requirements of e-business are more basic; to make the facilities of existing tried and trusted applications available to a wider reach.

Internal staff can be trained in the complexities of legacy systems. But staff within business partners, and even more so the general public, cannot be expected to use legacy systems unless accessing them is completely intuitive.

The ideal solution is to re-engineer legacy systems completely. But for many companies this is impracticable. It involves a large up front investment. It requires IT staff to become familiar with new technologies such as Java, or Microsoft's Com. And it involves discarding systems which work, where the business logic is sound, and where in many cases there has been recent substantial investment in ensuring Year 2000 compliance.

And for many mainframe users it is not desirable. 'Some customers would still rather code everything in Cobol,' says IBM IT Consultant Mike Silnickas.

So many mainframe sites do not want to start with a clean sheet of paper. For these, taking advantage of the opportunities offered by the internet means making processing logic and data that already exists to a much wider user base, and doing so in a way which makes it easy for them. There is no need or desire to alter the application; all that needs changing is the front end.

That is no different in principle from the problem that mainframe and mini users found at the start of the client-server revolution a decade ago. At that time, access to the host computer was typically via a character-based terminal - the green screen. The challenge was do away with character-based commands, and find a way of implementing the new breed of mouse controlled user interface, with windows, icons, and pull-down menus.

Companies then as now could completely re-engineer their applications, and many did. Some got what they wanted, but many spent more money than they needed to for the benefits they got. The downsizing craze of the early '90s proved an expensive diversion for many, and led to the trend for server and storage consolidation that still holds sway today.

Some users a decade ago were not convinced by the prevailing fashion. They spent a lot less money, kept their existing applications and IT infrastructure, and simply changed the front end. An industry sprang up to help them do this. Initially it was called terminal emulation, then PC connectivity.

That industry survives today, with the market leaders including Attachmate, NetManage, and WRQ. All these companies have adapted to the internet revolution, and now provide tools for web-enabling legacy applications. They are doing so with different approaches.

Attachmate describes itself as 'moving towards enterprise application integration'. The company started by selling basic terminal emulation products, and still sells them today, though they now account for a relatively small part of the business. Attachmate has progressed from there. According to business development manager Jon Newland: 'What we're doing now is allowing you to get information from a host without people knowing you're doing that.'

From a technical perspective, this has been a technical progression up the seven-layer ISO communications model. Says Newland: 'When we started we worked at the physical and link layers of the OSI seven layer model. Then we moved up a couple of layers to the networking/connectivity piece. Now we are into the session control and presentation layers, and later in the year we will move up into the application layer. We've done integration from day one, but we've gone up the ladder. Anybody who's got a host, and wants any form of integration, we can do.'

Newland sees Attachmate as helping the IT manager to retain control over the IT resource at a time when the arrival of the internet is threatening to take that control away. 'People talk about web-to-host, but what we do is host-to-web - opening up the mainframe to the web.

The mechanism is the same, but the approach is very different. In this approach the IT manager sees himself as a powerful gatekeeper.'

WRQ, similarly, is expanding from terminal emulation to a higher infrastructural level. The company has introduced a product called Reflection for the web, which provides Java emulation for mainframe, AS/400, Unix, VMS, and MPE systems.

According to Rob Andrews, Solutions Specialist with WRQ's UK distributor Wick Hill, Reflection 'also offers rejuvenation - on the fly you take a mainframe screen and make it look like a Windows application. That is in between full transformation and terminal emulation; it's still screen based, but you can add graphics, rearrange fields, and so on.'

This product is designed to help people keep the screens they are familiar with, rather than opening up applications to a wider audience. WRQ has introduced another product for this purpose, called Versastream Host Integrator (VHI).

VHI acts as an intermediary server between a legacy machine, such as a mainframe, and a web application server. It creates multiple Telnet sessions into the mainframe, and presents outputs from it to the web server as Java or Com objects.

Whereas Attachmate and WRQ have adopted their existing product lines to a new set of requirements, NetManage is transforming itself into a new services based company.

The company identifies three levels of web-legacy integration. First, making legacy applications look like a web site, for ease of access. Second, combining the facilities of several back-end systems, and presenting them as one. Third, adding new business logic to such a combination.

The first of these, says senior vice president for strategic development Peter Havart-Simkin, 'is the easiest one for connectivity suppliers to play in'. To be able to do the second and third, 'you need a family of slick development tools, and you also need a service/consultancy organisation to back it up.' NetManage has acquired these capabilities by a series of acquisitions, and has implemented them in a product set called OnWeb.

All very different
Now 'we've shifted from box-shifting to services, from $400 a product and declining to $250,000 a contract and going up. We've changed our way of selling, and the kind of people we employ to sell. The products are very different.'

NetManage now sees its opposition not as the traditional terminal emulation companies, but systems integrators, such as BEA Systems, Neon Systems, and even IBM, though Havart-Simkin is keen to stress there is no direct competition with Big Blue. 'We're providing WebSphere capability for the mid-tier market. You don't have to rewrite in Java; you don't have to rewrite any existing things at all.'

But despite all the changes, NetManage believes it is still very much in the same business it has always been. 'At heart it requires the same technology, the same access principles, the same mission-criticality,' says Havart-Simkin.

So when should you consider a simple front-end re-design, and when a more radical change? According to Attachmate's Jon Newland: 'If the business rules were OK when we did things the traditional way, and if the applications in the back office are still relevant, the best bet now is to integrate the back office into the front office. If they're not relevant, there is a real case to move that application off your system on to a new platform.'

Case study: The Brick
Canadian furniture and appliance retailer The Brick provides an example of web-enabling an IBM 3270 terminal system which has increased sales by 3 per cent to 5 per cent a month across 50 sales locations. It involved development of a web interface common to all business partners, and adding data mining capability.

The problem with things as they were lay in the credit approval system, which was cumbrous and slow, and was frustrating customers.

Brick sales representatives had to enter customer information into different lender mainframes using green screens and transmitting via remote 3270 gateways - a typical mainframe scenario. It meant a wait of up to 20 minutes for credit approval, plus a repeat process at a different lender if the first rejected the application. Each lender used by The Brick had its own individual 3270 interface, with its own codes and command lines, so the sales rep training requirement was heavy.

Using NetManage's OnWeb software, The Brick built a new application combining the mainframe data with a new web based front end.

The resulting improvement in sales has totalled over $1m a month. The company calculates that this paid for the development costs of the new system in just two days.

Web-enabling legacy systems is for many users a realistic alternative to producing new e-business systems from scratch. It offers the benefits of making applications and data available to a wider range of users while capitalising on existing investment. The traditional PC connectivity companies have developed products that allow web-enablement and are offering different levels of sophistication and cost.

Read more on Business applications