There are various ways to achieve what you say is good business sense.
BT only needed to secure a backup supplier, they didnt need to actually commit to deploying units from the backup provider to have that redundancy.
Also if you going to dual supply its common sense to at least have a comparative technical specification, not an entirely different chipset, and one that has a poor reputation to goto boot.
They could have got a backup supplier that also used broadcom. But I can tell you why they didnt, cost control. My speculation is that if the ECI problems were not discovered when they were (lack of forward vectoring capability) then I think the full last 2/3 of the rollout including BDUK would have been ECI, I expect they only switched back to hauwei for the final third because that was when they started considering vectoring and a realisation occurred it wouldnt work on ECI, and of course with vectoring actually been used in BDUK areas, those areas had to be hauwei.
I also expect on a higher value contract BT may have been able to agree a hardware upgrade program with ECI e.g. switching out M41s for V41s without paying for V41s at full market value, but again I speculate the prime driver for the contract was cost.
BT are not innocent of blame simply because its all too easy to blame ECI. If ECI breached their contract then BT would likely be sueing them, I expect the contract has been fulfilled to the specifications agreed to.
Dual vendor inevitably adds long term costs as staff have to be trained on two difference vendors and things like maintenance, feature upgrades and so on become more complicated.
There is nothing wrong with negotiating a second contract with someone else to keep the primary vendor on their toes, the mistake BT made was actively deploying the two different chipsets in the field for the same service. Broadcom is the biggest name in DSL and the entire rollout should have just been that chipset, it seems on g.fast BT have at least made sure they only having one chipset this time round, and thought better of committing to that new name that entered the g.fast market.
Also the redundancy means nothing if the company (BT) is not willing to swap out chips to fix problems, so in this case ECI not working optimally but no willingness to switch the vendors at the sites. In reality having two vendors in the field just doubles the chance of a problem.
Edited by Chrysalis (Wed 08-Feb-17 17:50:03)