Big Bang is certainly the way to describe it: At the Mark Logic London User Group, Freddie Quek of Wiley described how Wiley achieved their “big bang”, the project to redevelop what had been known as Wiley InterScience and created the Wiley Online Library. It was a thrilling tale, delivered with all the excitement of an explorer discovering the source of the Nile, or a pioneer crossing North America. What he described was indeed an achievement, the redevelopment of Wiley’s biggest customer-facing application, which has impressive-sounding statistics such as 65 million page views per month, 200 million visitors annually, and 1,600 queries per second.
What were the lessons that came out of this project?
- The benefits of prototyping. One of the team put together a prototype using Mark Logic in four days, and this enabled the idea to be demonstrated. This is because, as Freddie Quek explained, however carefullly you capture the requirements, the users will always want something different, and you might as well find out as early as possible in the project, by showing them.
- One example of the benefits of prototyping was explained as “the little green icon” - a button that displays in the search results to show if the content is viewable without payment or restricted to subscribers only. Apparently the prototype fixed in a few hours something that using the old code base would have taken nine to twelve months to develop. I regard these sort of statistics as “magic”- since we aren’t told how they are achieved, they might as well be magic.
- Sometimes changing everything (the “big bang”) is preferable to incremental change. In this case, the big bang comprised a new code base 9(they switched from C++ to Java), a new content model, a new content repository, and a new search engine (about as big a change as you can make in one go). The decision was taken to change everything because the legacy system was “monolithic”, which he explained as “everything mixed up with everything else”- the user registration mixed up with the search engine, which was mixed up with the e-commerce. Interestingly, there was a question from the audience on this point about the implication that the new Mark-Logic based system was any less monolithic than the one it had replaced. It appears this problem still exists, because the next phase of Wiley’s platform development is looking to modularise the code into separate groups.
- This meant teaching the team of developers an entirely new programming language Wiley brought in two groups of external teams: first, MarkLogic developers (in some cases persuaded to relocate from the US), and secondly, evangelists whose job it was to help carry the message across Wiley.
Equally interesting, for me, was what he would have done differently, if he had the chance of doing the project again. His response was: Try not to be the first mover. I wish there had been a community, people I could talk to, to reduce the risk.
But as we all know, there are times when a decision has to be taken, and the absence of any precedent or parallel initiatives does not make it any less important to proceed.