This is a story of a fantasy company situation but most likely very similar to yours
So I am sure you will recognize yourself and your company
The story is a about how IT is effected by company policies
The story is also about what we can do in IT to mitigate the company policies with new technologies and DevOps
Form your future
Presenters notes has been added as lines in the presentation
FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756
It creating the future_fantacyedition
1. Create the future a story by Max Börebäck
• This is a story of a fantasy company situation but most likely very similar to yours
– So I am sure you will recognize yourself and your company
• The story is a about how IT is effected by company policies
• The story is also about what we can do in IT to mitigate the company policies with new
technologies and DevOps
• Form your future
2. • Infrastructure Architect for Java applications
• Newly trained Scrum Master
• Automation in focus
• Focused on improvement
• Mixed background;
– Developer
– Project manager
– Process manager
– Team manager
– Technical expert
– Monitoring (event management)
• Family wife and 4 kids
• Living in Tullinge south of Stockholm
• Spend vacations in Torrevieja in Spain
Literature I recommend
• The Goal
– by Eliyahu M. Goldratt
• It’s not luck
– by Eliyahu M. Goldratt
• The Phoenix project
– by Gene Kim, Kevin Behr, George Spafford
• Scrum: The art of doing twice the work in
half the time
– by Jeff Sutherland
• The Power of Ambition: Unleashing the
Conquering Drive Within You!
– by Jim Rohn
Who is Max?
6. Presenter notes for previous slide
• The company has a Goal
• Make money now and in the future
We have also 3 critical success factors that helps us reach the goal
• Maximize throughput (Get as much as you can out to the customers so you get in money for
your services)
• Minimize inventory cost (inventory cost money, so we should not stock anything, but get it out to
the customer)
• Minimize operations expenses (utilize the personal so we are as productive as we can)
The company also has decided on the strategy
• Shall be in the Service sector
• Business systems for you
• Shall have digitalized services
• Have our own IT organisation
So when it coms to organising our company we are setting up our IT Organisation and decides on
how it shall look like
We divide it into to main departments Development and Operations
When we do this we are also setting up a small fence between the departments as we tell them to
focus on different things
6
8. Presenter notes for previous slide
• Our company has a goal, a strategy and an it organisation
• The company now decides that we need to measure how good our IT organisation is and we
have the two departments Development and Operations
• So we put in policy's for each department
• For development one of the measurement points is how fast we can develop new functionality
• For Operations we put in measurements on availability and operational expenses
• What we have done now is that we have converted the fence into a wall between the IT
departments as thy are now measured and driven by different policys
8
10. Presenter notes for previous slide
• The IT department also has to have its Goal
• As the intention for the company is to have Digitalized services and we have taken the strategy
to provide them with the help of our own IT organization so we inherit the strategy as our goal
• The Goal is Digitalized services
• We also have to to have critical success factors that helps us reach the goal, and to start with
we inherit the companies critical success factors that we then can complement with some more
• Our critical success factors are
• Maximize throughput (Get as much as you can out to the customers so you get in money for
your services)
• Minimize inventory cost (inventory cost money, so we should not stock anything, but get it out to
the customer)
• Minimize operations expenses (utilize the personal so we are as productive as we can)
• I will add in some more critical success factors later in the slide, you will see them in the goal
tree
10
12. Presenter notes for previous slide
• Ok so lets have a look at our IT factory
• We have our goal and success factors that shall help us run the factory in a good way
• Development is developing new software and solutions
• We we developing new things we are adding a new software into our software inventory
• All inventories is holds inventory cost as we have put in time and money into it
• This is an investment
• And an investment can be calculated in money days
• And to get return of the investment you need to sell it and when more money days has come
into the company than the money days spend on development and operations we get an return
12
20. Presenter notes for previous slide
• Our policy's are impacting us
• Policy for development is to provide new functionality as fast as possible
• Policy for operations (production) is stability and operational expeses
• In this picture we have a well balanced factory, we develop, we put in to production, we cell
and customers are buying the same amount we develop so all we develop are generating
money back
• So no overhead
• But how does the policies impact us?
20
22. Presenter notes for previous slide
• Our policies have created a conflict between development and operations
• Our goal is the same (objective) we shall deliver digitalized services
• So in order to provide competitive digitalized services we need to provide new value added
fatures as fast as possible (TTM) AND we need to have stable operations
• In order to provide new value added features as fast as possible (TTM) we must change
often
In order to provide stable operations we must change less often
And we can not do both
So how shall we solve this conflict?
Lets look again on the policies the company has put on us and see if we can do something
22
23. 23
IT
Factory
Develop Production Sales Customer
SW
$
SW
$
SW
$
Measurement
Policy
Measurement
Policy
Local Optimization
SW
$Agile
CI
CD
SW
$
SW
$
SW
$
Maximize
throughput
Minimize
inventory
cost
Minimize
operations
expenses
Digitalized
services
24. Presenter notes for previous slide
• We have a conflict that has its origin in the policies for how we shall measure our IT departments
• Lets see the effect of that
• We have a balanced production pipeline
• What we develop, we put in production , we sell it and our customers buys it
• Our policy that drives Development is pushing them to provide more and faster
• So Development is adopting the latest methods in being more productive
• Development is adopting agile project methods, and new methods for developing like Continues Integration and Continues
Delivery
• Development can then in this example double there capacity
• But Operations have different objectives as they are driven by being conservative and change as seldom as they can to
preserve stability
• So production is not increasing its speed
• So even if sales can sell what is developed and we have a market that are willing to pay for the new things we can not deliver
• So the only thing that happens is that our Inventory Cost is going up
• So we are preventing our self to meet the goal
• In systems thinking this is called Local Optimization
• Critical Chain is Constraint Management and there is always one and only one link at any point in time that is your weakest
your current bottleneck
24
25. What can we do?
• Focus on customer value
• Be smarter
• Avoid local sub optimization
• Optimize the whole system
• Remove your constraints (bottle neck)
• Work together
• Get rid of waste
25
Develop Production Sales Customer
26. Presenter notes for previous slide
• Break the policies and put in policies that help you meet the goal for your IT Organisation as a
whole and your company
• Start working together, and be open there is always two perspective like in the picture are there
3 or 4 or both
• Start communicating, talk talk talk
• Work together
• And get rid of waste
• Common waste is wait time, in your TTM pipeline you have a lot of wait time
• Waiting for test to be done
• Waiting for other teams
• Waiting …
You are also dubbel working teams are inveting the same wheal
Teams are doing what another team is sopose to do
26
27. What do we need
• Automation
– Automation for building
– Automation for testing
– Automation for deployment
• Instances on demand (static or dynamic) (linux servers and docker containers)
– Decommission unused instances to free up inventory to other projects
• 0 downtime for new application versions
• Canary deployment (Piloting group for new versions)
• Transparent user transitions from one version to another
• Checkpoints / tollgate in TTM Pipeline
• Control
– Monitoring
– Dashboard
• Status and overview
• Capacity analyse
• Performance analyse
– Log file management
• Services definition
– Value services for customers
– IT services and sub services
27
28. New instance in < 20 minutes
28
Production App
cloud
App v 1 App v 1
Project cloud
App v 2
Dashboard and control center
Central logs
App v 2
29. Presenter notes for previous slide
• Why not be able to put a new instance of the new version of the software in production within
20 minutes after it is tested?
• This picture illustrates a scenario
• We have a test system cloud where we very fast can spin up new instances for a project so
they can do development testing, functional testing, performance testing, ui testing etc etc.
• When ready push it into production together with the existing version
• Canary deployment
• Give a selected group of your customer access to the new version, you will now have a
production quality test done
• When its done scale up the new version and scale down the old version
• You provide 0 downtime to end users
• Monitor everything
• Star monitoring in development and continue doing this in all environments
• Manage your logs in a good way
29
31. Presenter notes for previous slide
• Analyse what you do to find out what you need and what you should stop doing
• There is a good tool for this called Logical Thinking Process tools that is designed by Goldratt and has now evelved to what they are today
• Here is an example of how one part of the tool can be used to identify what you need and why you need it, its called Goal Tree
• I have taken the 3 critical success factors from our IT organisations goal and added a new one from the development policys
• The goal tree is a necessary tree, so we start from the top and ask our self what are the last things that has to happe or exist for this to
happen
• So what has to happen or exist to maximize the throughput?
• In my tree I have put in two things
• Provide systems on demand
• Release more often
• There is more things but you shall also make you definition without detailes
• And the best is if the are full sentences so its clear
• Building this tree will help you identify all the things you need to have
• You can then match that to what you have and get rid of the things not needed and create the things you realy need
• This will help you break down the walls between your departments
• This is also helping you justify what you need to do to the company steering group
• In this tree you can also see relations between our Critical Success Factors
• So if we continue breaking down Release more often
• To do that we must minimize production errors
• Which is a shared item with another critical success factor
• I do recomend reading the book the Goal and for you that likes to learn more Dettmer has written a good book in Logical Thinking Process
31
32. Docker VS vm’s
• Java Application servers cost money
• Java Applications are memory hungry
• Get rid of waste!
• Application needs 1 GB of memory
• OS needs 1 GB of memory
• Waste = 1 GB of memory
• Operations cost, patching, maintenance
• Waste = OS operational cost
• Docker do not remove all Waste ! But it is lowering the waste for application operation
• Docker containers can be packaged by developer
• Give consistent testing trough out the whole TTM pipeline
32
Minimize
operations
expenses
Minimize
inventory
cost
33. Presenter notes for previous slide
• In the Goal tree we could see that Docker is one thing we need
• If we look in the java area memory is one of the things that is consumed most, application in
general eats more memory than cpu
• An estimation is that you can pack 30% more application on a HW server with docker
containers compared to if you use vmware instances
• There is also a maintenance cost that will be saved
• But do not foul yourself docker is not free it also requires operations
• And be careful select the platform that will work best for you or build it on your own, but I
recommend that you select a ready platform for you to run on prem as well as scale out on
private and public cloud
33
34. What about ITIL
• We need to monitor to catch and feed back information to development (incidents)
• We need root cause analyses if incidents repeat or are sever (problems)
• We need control of what is out in production and when it happens (change)
• We need minimize inventory cost (capacity planning)
• We need to maximize TTM (Service transition)
• We need … ITIL processes
• We need to change how we work with the processes and how we hook them into Agile teams
• Change management has to be fast
34
Maximize
throughput
Minimize
inventory
cost
Minimize
operations
expenses
35. Presenter notes for previous slide
• This picture is about all the talks about DevOps and that it will make ITIL obsolete.
• It will not, ITIL is a good process toolbox and use the goodies from it
• And be practical and make the processes work well with your demand on TTM and way of
working with containers
• In general to make it dynamic you need to be able to spin up new instances fast that will live
from a few minutes to months
• And you need to be practical in your way in handling changes
• Utilize auto aprovement by setting tollgates and measurements
35
39. Presenter notes for previous slide
• TTM Pipeline this is our factory pipeline, I have left out sales etc.
• This picture illustrates a pipeline from development to production
• In the pipeline I have put in traffic light, if you get green you pass on to next step in the TTM Pipeline if you get ref you have to improve
• The more robust verifications you have and the more you automate them also makes it more possible to increase the number of production setting you can do.
• You have to decide on how often you want to go to production
• Multiple times per hour
• Multiple times per day
• Multiple times per week, month year
• The number of times to go to production defines how long time you have for each step in the deployment pipeline
• If you have a pilot production environment for canary deployment tests, you have to have enough time there for users to verify the system.
• With clear criteria for each traffic light makes it very clear to everyone what failed so you can put the efforts on the right thing.
• You do not need automation to establish the TTM Pipeline with traffic light
• Go ahead and do it right away it will help you with many things
• By defining your TTM Pipeline you will identify all the stations in the pipeline and will be able to start monitor how long time each station takes and the wait time between
stations
• And also the quality, if you find production errors, which traffic light was failing
• You will very soon see that you are working with continues improvement
• WHY should be your favourite word
• Why are we waiting
• Why did we not detect the error in our traffic lights
• Why
• Why
• Why
39
40. What shall we focus on now? Output of the system!
40
Critical Chain
Goal Tree analyse
Based on
Necessary Conditions
Or
a full LTP analyse & plan
41. Presenter notes for previous slide
• Your company has stated the goal and the critical success factors
• We shall earn money
• We shall have high throughput
• We shall minimize our expenses
• Utilize the TTM Pipeline to make this happen
• Identify your stations in the pipeline and start measure to identify where you have your bottle necks your
constrains
• Dive in to each station and break it down to your Station Pipeline
• Identify your Station Pipeline Constraint (bottle neck)
• Use the Goal Tree to help you identify what you need and why you need it
• Do not do things just because you always have done them create the tree and visualize what you realy need
• Make every one understand your tree so you can create a momentum in your Continues Improvement of
your TTM Pipeline
• Do not forget that the world is changing, new demands, so you have to continuously review your Goal Tree
and adjust it to the reality and the demands of your reality.
• If the demand for your services are going down, make sure you adjust your strategy and tactical decision so
you can continue meeting the goal it might be that you have to change direction completely
• Like Toyota they did not start as a car manufacturing company they ended up there as a result in a change in
demands and they adjusted there strategy to be successful.
41
43. Presenter notes for previous slide
• Communicate
• Talk with each other
• Make sure that everyone understands the needs each part of the organisation have and how
you can support each other
• Business has to talk with development but also with operations to secure that the whole
lifecycle of the service if covered
• So start talking and start working together
• Make your system work as as as system towards to systems Goal and Critical Success
Factors
• Remove your local optimization in favor for system optimizations
43
Hinweis der Redaktion
This is the house of the company, the boundary's of the system and what we control
The company has a Goal
Make money now and in the future
We have also 3 critical success factors that helps us reach the goal
Maximize throughput (Get as much as you can out to the customers so you get in money for your services)
Minimize inventory cost (inventory cost money, so we should not stock anything, but get it out to the customer)
Minimize operations expenses (utilize the personal so we are as productive as we can)
The company also has decided on the strategy
Shall be in the Service sector
Business systems for you
Shall have digitalized services
Have our own IT organisation
So when it coms to organising our company we are setting up our IT Organisation and decides on how it shall look like
We divide it into to main departments Development and Operations
When we do this we are also setting up a small fence between the departments as we tell them to focus on different things
Our company has a goal, a strategy and an it organisation
The company now decides that we need to measure how good our IT organisation is and we have the two departments Development and Operations
So we put in policy's for each department
For development one of the measurement points is how fast we can develop new functionality
For Operations we put in measurements on availability and operational expenses
What we have done now is that we have converted the fence into a wall between the IT departments as thy are now measured and driven by different policys
The IT department also has to have its Goal
As the intention for the company is to have Digitalized services and we have taken the strategy to provide them with the help of our own IT organization so we inherit the strategy as our goal
The Goal is Digitalized services
We also have to to have critical success factors that helps us reach the goal, and to start with we inherit the companies critical success factors that we then can complement with some more
Our critical success factors are
Maximize throughput (Get as much as you can out to the customers so you get in money for your services)
Minimize inventory cost (inventory cost money, so we should not stock anything, but get it out to the customer)
Minimize operations expenses (utilize the personal so we are as productive as we can)
I will add in some more critical success factors later in the slide, you will see them in the goal tree
Ok so lets have a look at our IT factory
We have our goal and success factors that shall help us run the factory in a good way
Development is developing new software and solutions
We we developing new things we are adding a new software into our software inventory
All inventories is holds inventory cost as we have put in time and money into it
This is an investment
And an investment can be calculated in money days
And to get return of the investment you need to sell it and when more money days has come into the company than the money days spend on development and operations we get an return
When we are finished with development we move it to the production and we still have the investment cost
We move it to sales and we still have the investment cost
Our customers buys it and bring in money to the company
Our policy's are impacting us
Policy for development is to provide new functionality as fast as possible
Policy for operations (production) is stability and operational expeses
In this picture we have a well balanced factory, we develop, we put in to production, we cell and customers are buying the same amount we develop so all we develop are generating money back
So no overhead
But how does the policies impact us?
Our policies have created a conflict between development and operations
Our goal is the same (objective) we shall deliver digitalized services
So in order to provide competitive digitalized services we need to provide new value added fatures as fast as possible (TTM) AND we need to have stable operations
In order to provide new value added features as fast as possible (TTM) we must change often
In order to provide stable operations we must change less often
And we can not do both
So how shall we solve this conflict?
Lets look again on the policies the company has put on us and see if we can do something
We have a conflict that has its origin in the policies for how we shall measure our IT departments
Lets see the effect of that
We have a balanced production pipeline
What we develop, we put in production , we sell it and our customers buys it
Our policy that drives Development is pushing them to provide more and faster
So Development is adopting the latest methods in being more productive
Development is adopting agile project methods, and new methods for developing like Continues Integration and Continues Delivery
Development can then in this example double there capacity
But Operations have different objectives as they are driven by being conservative and change as seldom as they can to preserve stability
So production is not increasing its speed
So even if sales can sell what is developed and we have a market that are willing to pay for the new things we can not deliver
So the only thing that happens is that our Inventory Cost is going up
So we are preventing our self to meet the goal
In systems thinking this is called Local Optimization
Critical Chain is Constraint Management and there is always one and only one link at any point in time that is your weakest your current bottleneck
Break the policies and put in policies that help you meet the goal for your IT Organisation as a whole and your company
Start working together, and be open there is always two perspective like in the picture are there 3 or 4 or both
Start communicating, talk talk talk
Work together
And get rid of waste
Common waste is wait time, in your TTM pipeline you have a lot of wait time
Waiting for test to be done
Waiting for other teams
Waiting …
You are also dubbel working teams are inveting the same wheal
Teams are doing what another team is sopose to do
Here is some of the things we need
Why not be able to put a new instance of the new version of the software in production within 20 minutes after it is tested?
This picture illustrates a scenario
We have a test system cloud where we very fast can spin up new instances for a project so they can do development testing, functional testing, performance testing, ui testing etc etc.
When ready push it into production together with the existing version
Canary deployment
Give a selected group of your customer access to the new version, you will now have a production quality test done
When its done scale up the new version and scale down the old version
You provide 0 downtime to end users
Monitor everything
Star monitoring in development and continue doing this in all environments
Manage your logs in a good way
Analyse what you do to find out what you need and what you should stop doing
There is a good tool for this called Logical Thinking Process tools that is designed by Goldratt and has now evelved to what they are today
Here is an example of how one part of the tool can be used to identify what you need and why you need it, its called Goal Tree
I have taken the 3 critical success factors from our IT organisations goal and added a new one from the development policys
The goal tree is a necessary tree, so we start from the top and ask our self what are the last things that has to happe or exist for this to happen
So what has to happen or exist to maximize the throughput?
In my tree I have put in two things
Provide systems on demand
Release more often
There is more things but you shall also make you definition without detailes
And the best is if the are full sentences so its clear
Building this tree will help you identify all the things you need to have
You can then match that to what you have and get rid of the things not needed and create the things you realy need
This will help you break down the walls between your departments
This is also helping you justify what you need to do to the company steering group
In this tree you can also see relations between our Critical Success Factors
So if we continue breaking down Release more often
To do that we must minimize production errors
Which is a shared item with another critical success factor
I do recomend reading the book the Goal and for you that likes to learn more Dettmer has written a good book in Logical Thinking Process
In the Goal tree we could see that Docker is one thing we need
If we look in the java area memory is one of the things that is consumed most, application in general eats more memory than cpu
An estimation is that you can pack 30% more application on a HW server with docker containers compared to if you use vmware instances
There is also a maintenance cost that will be saved
But do not foul yourself docker is not free it also requires operations
And be careful select the platform that will work best for you or build it on your own, but I recommend that you select a ready platform for you to run on prem as well as scale out on private and public cloud
This picture is about all the talks about DevOps and that it will make ITIL obsolete.
It will not, ITIL is a good process toolbox and use the goodies from it
And be practical and make the processes work well with your demand on TTM and way of working with containers
In general to make it dynamic you need to be able to spin up new instances fast that will live from a few minutes to months
And you need to be practical in your way in handling changes
Utilize auto aprovement by setting tollgates and measurements
Going agile and DevOps do not change the need for solid ITIL processes
TTM Pipeline this is our factory pipeline, I have left out sales etc.
This picture illustrates a pipeline from development to production
In the pipeline I have put in traffic light, if you get green you pass on to next step in the TTM Pipeline if you get ref you have to improve
The more robust verifications you have and the more you automate them also makes it more possible to increase the number of production setting you can do.
You have to decide on how often you want to go to production
Multiple times per hour
Multiple times per day
Multiple times per week, month year
The number of times to go to production defines how long time you have for each step in the deployment pipeline
If you have a pilot production environment for canary deployment tests, you have to have enough time there for users to verify the system.
With clear criteria for each traffic light makes it very clear to everyone what failed so you can put the efforts on the right thing.
You do not need automation to establish the TTM Pipeline with traffic light
Go ahead and do it right away it will help you with many things
By defining your TTM Pipeline you will identify all the stations in the pipeline and will be able to start monitor how long time each station takes and the wait time between stations
And also the quality, if you find production errors, which traffic light was failing
You will very soon see that you are working with continues improvement
WHY should be your favourite word
Why are we waiting
Why did we not detect the error in our traffic lights
Why
Why
Why
Your company has stated the goal and the critical success factors
We shall earn money
We shall have high throughput
We shall minimize our expenses
Utilize the TTM Pipeline to make this happen
Identify your stations in the pipeline and start measure to identify where you have your bottle necks your constrains
Dive in to each station and break it down to your Station Pipeline
Identify your Station Pipeline Constraint (bottle neck)
Use the Goal Tree to help you identify what you need and why you need it
Do not do things just because you always have done them create the tree and visualize what you realy need
Make every one understand your tree so you can create a momentum in your Continues Improvement of your TTM Pipeline
Do not forget that the world is changing, new demands, so you have to continuously review your Goal Tree and adjust it to the reality and the demands of your reality.
If the demand for your services are going down, make sure you adjust your strategy and tactical decision so you can continue meeting the goal it might be that you have to change direction completely
Like Toyota they did not start as a car manufacturing company they ended up there as a result in a change in demands and they adjusted there strategy to be successful.
Communicate
Talk with each other
Make sure that everyone understands the needs each part of the organisation have and how you can support each other
Business has to talk with development but also with operations to secure that the whole lifecycle of the service if covered
So start talking and start working together
Make your system work as as as system towards to systems Goal and Critical Success Factors
Remove your local optimization in favor for system optimizations