All too familiar
When I was in ICL, the call-logging-ticketting system (SIAM) ran on the old mainframes. It was old, clunky, had some restrictions, but worked.
It was due for an upgrade, but the suits made a deal for some off-the-shelf-but-never-heard-of replacement that simply didn't do what was required. Actual users of the old service obviously hadn't been contacted about requirements.
After much moaning, and bespoke code changes that didn't help, along with general reliability issues, it was eventually dumped for another off-the-shelf system (and the ICL guy responsible for the contract left to join the board of the company that created the software.)
The replacement, whilst better, was still missing stuff and had to be tweaked for some time (ICL was big enough that they could demand tweaks)
________________________
This one is even worse, and more relevant to the article.
I worked on the team that provided ASPECT - a monitoring system for the Unix servers (and some forms of VME monitoring), and collated messages that could be directed to pagers/call systems/mobile phones etc.
Pretty standard stuff these days, but this stuff had been around for years - long before I joined the team.. Many people from council IT workers to NTL/Virgin/ICL/DOD staff would have been aware of ASPECT.
Near the end of the 90's, corporate announced this "brilliant deal" with CA to rollout their monitoring tools company-wide. Whoever had made that deal had not researched what was currently being used, and didn't realize there was a whole team that wrote customized software to do the very same thing.
It was crud. The Unix modules were pants. They did provide a windows module, which was something we had needed to add, but damn was the overall thing expensive, and not suitable for our requirements (even other teams we had no direct relationship with told us it didn't fit their needs like ASPECT did).
Partly because of this, and partly because ICL had this bloody system in place already, I was tasked with writing stuff that would allow ASPECT backbends to interface with these clients.
Not long into this project, my suspicions were confirmed. I had to go to some internal conference thingie off site, and got talking to one bigwig about the frustrations with all the extra work we had. No surprise, he hadn't heard about ASPECT, and had been told that the CA deal would bring all these cool features we didn't already have. I explained that all the companies UK paging and alerts went through ASPECT, and mentioned some of the other things it did (some of which he was familiar with [like automated paging], but didn't know how it had all worked.)
He said he'd explore ASPECT to make sure it was used to its full potential, and would raise the fact he hadn't been told about it with the people associated with the deal.
His main feelings at the end of the conversation were something like "We have something, written internally, owned by us, that does the same thing, but is not generic, and works well with our specific use case, is run on a comparative shoestring budget... Why didn't I know of this before this deal happened?"
Well exactly.
I left the company a year or so later (2001ish) not long before ICL was swallowed whole into Fujitsu.
P.S. I also admit some of the ASPECT code was cruddy. During an audit I found some software written by someone I'd never met (who had long since left the company) even had this sprinkled through his C source:
system ("sleep 5");
