Computers and heat density

Alexey Toptygin alexeyt at
Wed Aug 16 18:49:36 CDT 2006

On Sun, 13 Aug 2006, Jules Richardson wrote:

> Oh, there was certainly some good hardware around. Actually, for embedded 
> type systems like telecomms, I believe that they haven't seen anything like 
> the kind of bloat that's around for desktop / server systems.

I was a telecom test engineer at my last job, and the software quality was 
atrocious. I once got a nice christmas gift from the engineering 
department of a Canadian company for finding ~30 critical (fails and 
doesn't come back up without operator intervention) bugs in their SIGTRAN 
product in less than 2 months; this after their sales department convinced 
our management that the product was deployed in "dozens of sites".

Then there was the GSN we bought from an Indian branch of a big American 
defense contractor; once again "dozens" of prior deployments were 
promised, but the product would not interoperate with anything other than 
their (buggy) test software. Parts of this real-time system were written 
in embedded TCL! I still have my notes on how to construct SS7 frames that 
will crash their stack; they refused to fix these because "the far end 
shouldn't be sending malformed frames" :-) I quit that job 1.5 years after 
the GSN was acquired, and only maybe 80% of the worst bugs had been fixed 
yet, however we had already deployed the product to 2 of our own 

Our own flagship piece of software hit 100% CPU at around 30-45 calls per 
second on a PIII 700 M CPU because every call control action has to update 
the Call Detail Records, and I gather there was a single spinlock over the 
whole (Microsoft Access) DB.

I'm betting with my test experience, I could make better quality products 
than 90% of the telecom market with 1-2 handpicked RF engineers and 3-4 
systems programmers. Of course the sales would still all go to the 
politically connected companies selling drek, so we would never make a 
profit. :-/


More information about the cctech mailing list