This talk examines how what you share will define you. The act of monitoring and dashboarding can have a profound effect, good or bad - on the attitudes and culture of the teams involved. With supporting case studies this session will show how you to help make any team more effective
6. @spoole167@spoole167
This talk is a story about how tackling the
communications gap made things better
• How it changed the behavior of the people involved..
• How you too can take a leap and leave your silo
9. @spoole167@spoole167
Modernising an IT group
I took over a large dev focused European IT team
that provided development services to developers
A team that would have been classified as “Slow IT”.
My job - modernise the team.
Support “Fast IT” for cloud development teams
I learnt a lot.
Then I talked to customers etc. I learnt even more.
13. @spoole167@spoole167
IT Dev
I learnt – making IT teams read the Phoenix Project can be counter productive
“I liked the book up and ‘til the part
where they started to modernise”
“So you want me to put
myself out of a job?”
15. @spoole167@spoole167
IT
A contract realised in new forms of self-service
assets like PaaS or IaaS , Kubernetes, Docker etc
A contract that covers new availability
requirements & speed to market
32. @spoole167@spoole167
Within a year everyone has a dashboard
• Every dev team
• Every support team
• Even managers
• And Executives..
In fact we build a network of
monitors with raspberry pi’s
To run the dashboard systems.
And they are monitored too..
Now over 4 million
dashboard events per day
36. @spoole167@spoole167
Why did simple dashboards go viral?
• Every dev team
• Every support team
• Even managers
• And Executives..
Now over 4 million
dashboard events per day
41. @spoole167@spoole167
A battle ground
Executives want ’levers’ to pull
Endless reinvention as they look for new insights..
Middle management want to control the message
Reduce the time spent ‘wasted’ on responding
Since (the elephant in the room) we all know the data
is inaccurate, old and any predictions mostly useless
42. @spoole167@spoole167
Exec: Reduce IT spend
Make devs more productive
New Executive Dashboard
Display progress in moving local workload into the cloud
Development assertion This is much harder
than it looks. We don’t have the data AND
it will cost us more in the end
43. @spoole167@spoole167
New Executive Dashboard
IT Challenges included:
1. Not being able to associate local workloads with the correct teams
2. Not being able to find all the servers
3. Not being able to accurately determine performance characteristics
• And dev workload owners expected to end up with a worse
environment so dragged their feet in helping
44. @spoole167@spoole167
When you have to distill a common goal into
a dashboard
Just enough status
Just enough insight
Just enough
Clear emergency indicators
45. @spoole167@spoole167
You find common ground together
• 1 person from IT worked with development
teams to get their data reclassified
• 1 person is not a threat
• 1 person makes a human connection
• 1 person asked ‘why’ , ‘what do you want’
0
2
4
6
8
10
12
Team 1 Team 2 Team 3 Team 4
Un assessed On Prem Units Able to move moved
46. @spoole167@spoole167
Unexpected benefits
• Better predictions of cost of moving workloads to cloud +
classification = each team can realistically discuss the cost/benefit of
moving their workloads to the cloud with Executives
• Managers can justify budget increases more effectively –they can
show how their existing capacity is being used.
• IT support now have an improved understanding of what their clients
are using machines for.
47. @spoole167@spoole167
All teams now want and need
the data to be accurate and
visible
•Which lead to improved
communications
between teams and
reduced the shouting
49. @spoole167@spoole167
Service Availability – a tale of impedance missmatch
• How do you calculate service availability?
• What’s your unit of time?
• What’s your availability period - 9x5, 24x5,
24x7?
• For SaaS products it is 24x7 ( minus very rare
scheduled downtimes)
• But what does available mean?
50. @spoole167@spoole167
Development IT team
• Provides a critical service to developers
• Asked to provide 24x7 support
• Dev IT: “we already do that”
• Dev teams beg to differ – service has regular downtimes..
• Dev IT : “Our availability calculations don’t have to
include regular scheduled downtimes”
• Dev (and me) WTF?
• what about all these other unexpected downtimes?
• Dev IT : “They are caused by events beyond our control
and so are not included”
• Dev (and me) WTF?
51. @spoole167@spoole167
Insight - Iceberg syndrome
• Many IT groups revolve around the SME
• Items move from one to another to get
completed so very few people have a vision of
an IT service as perceived by the end user.
• Published data is more for the service
providers use than the end user
• End users think they are dealing with an IT
department – but in reality they are dealing
with a small handful of people who have
responsibility for a part of the service.
52. @spoole167@spoole167
Solution
• Have the IT team understand they are
responsible for reporting the availability of
their service whatever the cause of
downtime.
• Have the IT team spend time with their users
to see what can be done to mitigate the
impact of scheduled downtimes
• Defined what available meant (hint not just
pingable)
55. @spoole167@spoole167
Culture changes?
• Creating a better definition of ‘available’
meant the IT team learnt what was important
to their end user
• Dashboards with ‘real’ status means that end
users don’t have to try to get answers via
ticket or phone - less stress. More trust.
• end users see the truth and can trust the IT
team more.
60. @spoole167@spoole167
Build Server upgrades
• A chore.
• Several hundred machines needed urgent security patching
• mostly by hand! ( a long story)
• How do you keep focus and motivate the teams who are doing this?
https://commons.wikimedia.org/wiki/File:US_Navy_090716-N-8590G-011_Members_of_the_damage_control_team_of_the_guided-
missile_frigate_USS_John_L._Hall_(FFG_32)_compete_in_the_pipe-patching_event_of_the_Damage_Control_Olympics.jpg
Navy Pipe Patching event
66. @spoole167@spoole167
Sort of
• Most teams now look at their dashboards
instead of monitoring emails, tickets etc.
• So do our clients
• So it’s more that we know together
67. @spoole167@spoole167
But most importantly..
Our customers know we know.
They understand how we work and they can see us working
Mutual trust through openness..
Teams feel more empowered, more in control of their work
Know they are doing the right thing
68. @spoole167@spoole167
So what have we learned?
That people
do change
1
Change comes
from mutual
understanding
2
By sharing
openly you’ll
learn more
3
All from moving
a little out of
our comfort
zones.
4