[LINK] Great bandwidth, poor latency

[Chris Johnson]

We should not be distracted into arguments about bandwidth that ignore
the latency, and forget the convenience of access on-demand that
dominates human behaviours, rather than the apparent efficiency and low
cost per bit of queued batched delivery.
I was reminded of the difference between bandwidth and timely access to
data by this story on koala genome research which is moving its data
management into the cloud.
"The handing of the data is something we were really struggling with. We
were taxiing hard drives around Sydney. The koala genome project is an
international project and with collaborators from all the over the world
and we were posting data to them.'
"The problems facing the koala genome project are not unusual for
research programs."
in The Australian 26/4/16
Researchers discover benefits of freedom to fail in the cloud
(paywalled)
http://www.theaustralian.com.au/business/technology/researchers-discover-benefits-of-freedom-to-fail-in-the-cloud/news-story/4fc3929493d5c4be0da048a37e2d0f51

(The end of the article  attributes the value of cloud computing for
research with the ability to lease space and experiment. I would give
more value to the ability to share data widely around a distributed team
with very fast update, and for researchers to do experimental analyses
of ideas before the heat of striking inspiration particles dissipate)
-----
"Never underestimate the bandwidth of a truckload of tapes trundling
down the highway."
-- 
Chris Johnson, Hon AsPro Computer Science ANU