On many occasions in my prior life at Demon Internet we laughed sales people
out of meetings when they offered SLAs that were limited to the value of a
months service. But, in the end *all* the salepeople offered the same deal.
Until when SLAs come with a pay back greater than the cost of the contract,
and in fact cover consequential losses, most service providers will treat
the failure to deliver within the SLA as a risk associated with the service
and not something more serious.
However: Would you (or anyone in the group) be willing to pay a premium for that, and how much is a "real" SLA, one covering consequential losses, worth to you?
You misunderstand. Which operators will offer this (backed by some
underwritten insurance) in an effort to be better than the competition ?
You know, I keep looking at the subject of this thread and
wondering what Dave Rand is up to.
It's a great question. For an operator to offer such a contract, I would imagine these assumptions must hold true:
1) The operator has a profit motivation: to sell what you and others will buy
2) that you and others are willing to pay a premium for a service that includes consequential damages in the SLA over a service that does not
3) that consequential damages can be defined to mutual satisfaction
4) the operator has the ability to quantify what the premium should be (the statistics of reliability and service delivery economics)
5) Starting reliability plus the overhead burden of managing such an SLA is such that the calculated premium is both marketable and profitable
I might have missed a few, but it's a start - Is this how you see the problem?
Buying car insurance doesn't alter the chances of your car breaking
down on a long trip. Insurance doesn't alter the chances of something
bad happening. Driver training, seat belts, and spare tires are things
which can reduce risk.
A warranty is not a substitute for due dilegence.
How can you perform due dilegence on a carrier? With a car, you
can open the trunk and check the spare tire. What is the best way
to check a spare circuit?
To me, the diverse route *is* the safety belt. If it does not work when it
is supposed to, I (or my mourning relatives) will sue the car companies arse
off - if the attribution for the failure rests with a design of
Safety / Seat belts are not decoration - and neither should a diverse route
or protection circuit be. (Yes, my English can boldly go... but I am still
waiting for the coffee to finish.)
How do you know it's a <insert vendor> field engineer trying to fix a
flat tire? He keeps swapping tires till he finds the flat one...
Make sure your testing procedure *really* tests the actual spare circuit,
not something else's representation of one. I know one poor soul who
wrote software to monitor-via-ping the other end of a point-to-point,
then watched in horror as a co-worker accidentally removed the entire
DSU from the rack, unwilling to say "Wait, that's the wrong one" because
the monitor hadn't burped an error.
Guess who's network was basically a triangle?