2. “Don’t make it
any worse.”
A tale of regression testing,
misguided expectations and
logical fallacies…
3. Instability – it’s a thing
• Projects to
stabilise
• Update/upgrade
underlying
technology
• While
maintaining
“normal”
service
4. Context…
• Wobbly
• PCI
• POC as a £££
revenue stream
• New for some,
old for others
• New database/
binaries/
hardware.
5. Two statements…
• “Don’t make it any worse”
• “Customers shouldn’t be
able to tell the
difference”
6. Immediate Questions…
• “To determine if
something is
worse, I’ll need
to know how bad it
is?”
• “What happens if
its good for some
too?”
• “For customers to
notice, I’ll need
to know what they
pay attention to…”
7. Deeper Questions…
• “We’re going to change everything,
how can I test that nothing has
changed?”
• “If you change the responsiveness
and capacity of a system, that’s
still noticeable?”
8. Orders of Ignorance…
• Little
codified
knowledge.
• We wouldn’t
know how to
know if
something had
changed…
9. #1 Risk Modelling
• FIBLOTS again!
• 4 people &
whiteboard x n
• People were
the primary
oracle
• With a bit of
existing
system…
10. #2 Light the path…
• Map the system
• Explore the thing
• Decide what sheds light
14. Results - Everything
changes but you…
Client A - “This
thingy responds
much faster than
before…”
Client B – “Before
we only do X jobs
a minute, now its
increased to Y
jobs”
15. Results - Move any
mountain?
• When does one stop
exposing a previously
unseen mountain of
knowledge?