Modernising Change
Management
with Enterprise DevOps
Your hosts
Nigel Kersten,
Field CTO,
Puppet
Nick Moore,
Senior Technical
Account Manager,
Puppet
The primary
problem we see
over and over
again...
IT Changes are far too often like this:
• Big, heavy changes
• With lots of manually provided context
• Delivered into inconsistent environments
• With slow feedback cycles
• At every step in the process
• In a highly siloed organisation
• With poor communication
So what is
modernised
change
management
then?
The best change management is:
• Small batches delivered frequently
• Fast to deploy
• Fast to remediate
• Fast feedback cycles
• Driven by self-service
• Pervasive metrics
• Early stakeholder involvement
Which Enterprise
DevOps
principles help?
Which DevOps principles help here?
● Measurement and Systems Thinking
○ Examine and measure the performance of
the whole system
● Create and Amplify Feedback Loops
○ Give all participants a way to provide
feedback
○ Collect the feedback
○ Act upon it
● Continuous Learning
○ Re-examine processes for optimisation
opportunities
● Automation
○ Underpins everything...
Focus on
Automation for
Speed
Automation enables
the future of change
management
• Automated Provisioning
• Automated Configuration Management
• Automated Testing
• Automated Compliance
• Automated Security Policies
Yet automation alone is insufficient!
At the end of the day, it’s always
about improving how people
interact
Automation can serve to codify a
social contract between teams
Automation can serve to codify a
social contract between teams
Real World
Challenges
Slow and heavy
approval processes
• Often slow because approval requires
going up the management hierarchy
• Often heavy because the approver needs
to be provided with all the context
• Often contain unnecessary steps that
were instituted due to historical incidents
• IT request approvals are often a huge
problem here – firewall changes,
hardware, patching, procurement.
As a leader, resist the urge to put
yourself in the approval chain
As much as possible, approval
should be done by people who
already have the appropriate
context
Even better, drive your teams to
automate the verification
processes for approval!
Focus on IT Request Fulfilment
- it impacts everything
Lack of parity
between test and
prod environments
• Without automation, pre-production
rarely ever matches production
• If production is completely
inaccessible to the people developing
the software – parity is even more
critical!
• Long-lived shared testing
environments are a strong predictor of
failure
Use automation and
infrastructure-as-code to create
parity between production and
pre-production
Investing in realistic test data
is critical
Too many changes
require approval from
too few approvers
• Your developers will always
outnumber your approvers
• Your approvers will always err on the
side of caution - that’s their job!
Involve approvers and auditors in
designing automation - early in the
delivery lifecycle
Start with automating the inputs to
your manual change processes
1. Continuously Re-examine
processes for optimisation
2. Automate the software delivery
pipeline end to end - infrastructure
and apps
3. Push approvals down to the people
who already have the context
4. Automate approvals, compliance
and security policy controls
5. Encourage Collaboration between
approvers, auditors and IT early in
the delivery lifecycle
6. Optimize IT request fulfilment.
Q & A
Thanks!

Modernising Change Management with Enterprise DevOps

  • 1.
  • 2.
    Your hosts Nigel Kersten, FieldCTO, Puppet Nick Moore, Senior Technical Account Manager, Puppet
  • 3.
    The primary problem wesee over and over again...
  • 4.
    IT Changes arefar too often like this: • Big, heavy changes • With lots of manually provided context • Delivered into inconsistent environments • With slow feedback cycles • At every step in the process • In a highly siloed organisation • With poor communication
  • 5.
  • 6.
    The best changemanagement is: • Small batches delivered frequently • Fast to deploy • Fast to remediate • Fast feedback cycles • Driven by self-service • Pervasive metrics • Early stakeholder involvement
  • 7.
  • 8.
    Which DevOps principleshelp here? ● Measurement and Systems Thinking ○ Examine and measure the performance of the whole system ● Create and Amplify Feedback Loops ○ Give all participants a way to provide feedback ○ Collect the feedback ○ Act upon it ● Continuous Learning ○ Re-examine processes for optimisation opportunities ● Automation ○ Underpins everything...
  • 9.
  • 10.
    Automation enables the futureof change management • Automated Provisioning • Automated Configuration Management • Automated Testing • Automated Compliance • Automated Security Policies Yet automation alone is insufficient!
  • 11.
    At the endof the day, it’s always about improving how people interact
  • 12.
    Automation can serveto codify a social contract between teams
  • 13.
    Automation can serveto codify a social contract between teams
  • 14.
  • 15.
    Slow and heavy approvalprocesses • Often slow because approval requires going up the management hierarchy • Often heavy because the approver needs to be provided with all the context • Often contain unnecessary steps that were instituted due to historical incidents • IT request approvals are often a huge problem here – firewall changes, hardware, patching, procurement.
  • 16.
    As a leader,resist the urge to put yourself in the approval chain
  • 17.
    As much aspossible, approval should be done by people who already have the appropriate context
  • 18.
    Even better, driveyour teams to automate the verification processes for approval!
  • 19.
    Focus on ITRequest Fulfilment - it impacts everything
  • 20.
    Lack of parity betweentest and prod environments • Without automation, pre-production rarely ever matches production • If production is completely inaccessible to the people developing the software – parity is even more critical! • Long-lived shared testing environments are a strong predictor of failure
  • 21.
    Use automation and infrastructure-as-codeto create parity between production and pre-production
  • 22.
    Investing in realistictest data is critical
  • 23.
    Too many changes requireapproval from too few approvers • Your developers will always outnumber your approvers • Your approvers will always err on the side of caution - that’s their job!
  • 24.
    Involve approvers andauditors in designing automation - early in the delivery lifecycle
  • 25.
    Start with automatingthe inputs to your manual change processes
  • 26.
    1. Continuously Re-examine processesfor optimisation 2. Automate the software delivery pipeline end to end - infrastructure and apps 3. Push approvals down to the people who already have the context 4. Automate approvals, compliance and security policy controls 5. Encourage Collaboration between approvers, auditors and IT early in the delivery lifecycle 6. Optimize IT request fulfilment.
  • 27.
  • 28.