ToC in Software Engineering Adopting a culture of improvement for IT

www.devops.co.il Evgeny Zislis @kesor6

Dr. Eliyahu M. Goldratt

2011 Death of Eliyahu Goldratt 1997 Critical Chain Brief history of Theory of Constraints 1970 Production scheduling software OPT The Goal 1984 It's Not Luck 1994 1988 Approach named Theory of Constraints TOC service offers the Viable Vision 2005 2008 Isn't It Obvious TOC Handbook & The Choice 2009 2000

How can we increase the throughput of this system? Identifying the System Constraint

How to determine the new size for chosen section? Identifying the System Constraint

Did throughput increase? What other physical changes occured? Identifying the System Constraint

Would increasing any other section result in more throughput? Identifying the System Constraint

Theory of Constraints Constraint is anything that limits a system from achieving higher throughput versus its goal.

1. Identify the constraint 2. Decide how to exploit the constraint 3. Subordinate the system to the constraint 4. Elevate the constraint 5. If constraint broken - go back to step one WARNING: Do not allow inertia to cause a system constraint! ToC Five Focusing Steps

Buffer(s) Decides when to launch work into the system. ToC for Manufacturing - DBR Protection Sets the pace Drum Protection Finished work New work Rope Signals the release of work into the system. Bottleneck Non-bottlenecks Non-bottlenecks

1. Utilization and activation of a resource are not the same ○ Activation is what should be done Utilization is what can be done “100% utilization of a non-bottleneck is wasteful.” 2. The level of utilization of a non-bottleneck is determined not by its own potential but by some other constraint in the system. 10 Rules of OPT

3. An hour lost at a bottleneck is an hour lost for the total system. 4. An hour saved at a non-bottleneck is just a mirage. 10 Rules of OPT

ToC for Projects - CCPM Results: * on average; based on ~60 projects Duration of projects -30% Number of projects finished +70% Throughput +53%

PMI - Pulse of the Profession 2017 pmi.org/-/media/pmi/documents/public/pdf/learning/thought-leadership/pulse/pulse-of-the-profession-2017.pdf

3 3 1 2 ToC Projects Planning 2 4 6 12 months Late 1 2 3 1 2 9 months

100% 50% 10% % consumption of project buffer % of critical chain completion 10% 50% 100%

100% 50% 10% % consumption of project buffer % of critical chain completion 10% 50% 100% P34 P64 P34 P66 P30 P52 P21 P11 P25 P49 P72 P98

X X X X X

Logical Thinking

No content

No content

Four Pillars of Theory of Constraints Every conflict can be removed Every situation is simple People are good Never say I know

example for software

High IT Performer Medium IT Performer Low IT Performer Deployment frequency Multiple per day (on demand) Once per week to once per month Once per month to once every half year Lead time for changes Less than one hour One week to one month One month to half a year Mean time to restore Lett than one hour Less than one day Less than one day (lower on avg) Change fail rate 0 - 15% 31 - 45% 16 - 30% 2016 State of DevOps Report dvps.me/2016-state-of-devops-report

gain more capacity reduce unneccessary rework

No content

Returns possible from cost of unnecessary rework avoided

reinvest into more experiments

removed waste $100M 1% 1/3

Potential Value Added from reinvestment in new features

June 16, 2011. Velocity Conference ~15k deploys / day in May 2011 May 4, 2016. AWS Summit Stockholm ~50 million deploys / year ~136K deploys / day “Deployment every 11.6s, 1,079 max in one hour. 10,000 mean number of hosts per deployment, with 30,000 maximum” - Amazon.com youtu.be/dxk8b9rSKOo

from the 2015 state of devops report

focus on quality reduction in downtime and defects

No content

No content

lead time for changes How long is the delay from requesting a change until customer value is delivered? time to restore service How long does it take for an abnormal behavior in the system to be detected and restored to the normal standard agreed way of operation? release frequency How many changes and features are being released to production in a fixed period of time? change fail rate How often the system fails or service disruption occurs that originates from lack of quality in the source and required rework? Performance KPIs

Returns possible from cost of downtime avoided

add it all together

add it all together

? www.devops.co.il

ToC for Software Engineering offer ○ Assessment of IT performance with a before-and-after benchmark. ○ Process improvement consulting services provided by senior system architects. ○ Technical training and mentoring sessions for software engineers and operations. ○ Architecture and system design service for innovative products. ○ Workshops for hands-on learning of cloud and devops technologies.