Discussion:
OT: Cloud Computing - How it can go wrong in a very expensive way.
(too old to reply)
David Wade
2020-12-11 13:52:25 UTC
Permalink
Many companies seem to rely on not paying bills to "improve cash flow".
When I worked in HP they removed all the order pads in February and
March and you could not order anything.

This was stupid. How on earth were you supposed to finish a job if you
needed for example a box of floppy disks, or some blank CDs. Insane....

... of course if you move your data to the cloud you only own it if you
pay the bill, so that "trick" does not work. When I worked for a local
council we often had home workers unable to work because some one had
decided not to pay their phone bills.....

... and if you have a BIG system and something goes wrong you may end up
with a big bill...

https://www.theregister.com/2020/12/10/google_cloud_over_run/

.. these guys were lucky. I think usually you would have to pay....

Dave
Bill Gunshannon
2020-12-11 17:12:45 UTC
Permalink
Post by David Wade
... and if you have a BIG system and something goes wrong you may end
up with a big bill...
https://www.theregister.com/2020/12/10/google_cloud_over_run/
.. these guys were lucky. I think usually you would have to pay....
I would guess so.
They wrote a web crawler with a software bug that caused an
infinite recursion (it did not discard links to pages
already processed).
With utility computing someone has to pay. Either the
customer with the software bug pays or all customers pay
a higher rate to cover such incidents. The first seems more
fair to me.
Keeping in mind my recent comment about The Cloud being a bad idea,
how about this:
“Software you don’t own in your infrastructure is a risk,”
DeSantis said.

That's peter DeSantis of AWS. :-)

bill
Arne Vajhøj
2020-12-11 18:21:57 UTC
Permalink
Post by David Wade
... and if you have a BIG system and something goes wrong you may end
up with a big bill...
https://www.theregister.com/2020/12/10/google_cloud_over_run/
.. these guys were lucky. I think usually you would have to pay....
I would guess so.
They wrote a web crawler with a software bug that caused an
infinite recursion (it did not discard links to pages
already processed).
With utility computing someone has to pay. Either the
customer with the software bug pays or all customers pay
a higher rate to cover such incidents. The first seems more
fair to me.
Keeping in  mind my recent comment about The Cloud being a bad idea,
    “Software you don’t own in your infrastructure is a risk,”
     DeSantis said.
That's peter DeSantis of AWS.  :-)
Said in the context of them writing their own firmware
for their UPS'es.

That is not an approach I will recommend as
a general approach for everyone.

But Amazon overall got over 1 million employees and
estimates of AWS employees are in the 25000-40000 range.

Some things may make sense for a company that size.

Arne
Arne Vajhøj
2020-12-11 19:02:45 UTC
Permalink
Post by David Wade
... and if you have a BIG system and something goes wrong you may end up
with a big bill...
https://www.theregister.com/2020/12/10/google_cloud_over_run/
.. these guys were lucky. I think usually you would have to pay....
That budget limit should have been a mandatory limit not an advisory one.
And upgrading that free Firebase plan without his explicit consent
was outright wrong IMHO.
I also noticed the claims about it taking a day for the totals to update.
Question: If Google had tried to claim the money from him in the courts,
would they have lost because of these issues ?
That will probably depend on the jurisdiction, fine print in ToS etc..

But cloud business is B2B and in general requirements for protecting
customers against their mistakes are smaller for B2B than for B2C.

Arne

Loading...