I guess it is becoming a recurring theme in this blog, but so many of the ‘new’ ideas we have in our industry are not new ideas, but new implementations. The latest would be the Mashup trend.
In the Wall Street Journal today, there was an article about the Mashups sewing data together. Mashups, for the uninitiated, are a development method for combining web-based applications into one mashed-up application. Thus, information from one internal CRM site and an external site can be combined in one we application. It is kind of a cliff note version of a portal.
Rewind 17 years to 1990. The IBM mainframe is starting it’s slow death (slower actually than Jimmy Durante in ‘Its a Mad, Mad, Mad Mad World), client server (ala Powerbuilder) is taking hold in corporate IT. But there is a legacy of 3270 & 5250 (green screen text based monitors), applications running the real corporate systems. Along comes ‘Screen Scraping’ technology.
Similar to mashups, screen scraping would integrate somewhere between the application and UI layer in traditional architecture. The tool would read a screen through a terminal emulator on Windows and assign the fields to new maps. More interesting, was the ability to combine multiple applications in one Scraped Screen. For example, you could scrape the G/L systems and the Order Entry systems and create a single application view.
Well, screen scraping faded away with client server and the advent of web applications, but it probably would have died on it’s own for architectural reasons. Putting the workload of what is really data integration at the UI layer limited scalability and was flawed.
Mashups, although more mature, are the screen scraping of modern time. The more complex the mashups become the more you are pushing what really belongs in the data layer into a UI layer. This is not to say that mashups do not have a place, they do. But, as with most other fads, architecture and architectural compliance will eventually point to the right solution.
