Pokrewne
- Strona Główna
- Tom Clancy Jack Ryan, Jr. 05 Zwierzchnik [2014]
- 252317758 Curso de Processo Penal Fernando Capez 2014
- Magdalena Parys Magik (2014)
- Sagan Carl Kontakt (2)
- Jon Trace Tom Shaman 01 Spisek Wenecki
- Andrzej Sapkowski 1. Krew Elfow
- Eddings, Dav
- Jordan Robert Spustoszone Ziemie
- Cortazar Julio Gra w klasy (SCAN dal 1110)
- Morressy John Kedrigern i wilkolaki
- zanotowane.pl
- doc.pisz.pl
- pdf.pisz.pl
- wrobelek.opx.pl
[ Pobierz całość w formacie PDF ]
.23.Gary Klein, Streetlights and Shadows: Searching for the Keys to Adaptive Deci-sion Making (The MIT Press, 2009), p.13.24.Kahneman and Klein, Conditions for Intuitive Expertise. 68 THE LEAN MINDSETResiliencyWhat does developing expertise look like in practice? A good exam-ple can be found in the discussion Resilience Engineering: Learningto Embrace Failure. 25 In this article, Jesse Robbins discusses how hisfirefighter training led him to convince his company (Amazon.com)that the best way to keep their complex systems up and running wasnot to make them perfect, but to make them resilient.In order to dothis, he started GameDay events events that are triggered by inten-tionally pulling the plug on a data center (or a similar self-inflictedfailure).For a typical GameDay event, it may take dozens of peopletwo or three days to get the system back to normal.Once things areworking again, a detailed review occurs in which dependencies andprocess weaknesses exposed by the exercise are documented and mostof them are rapidly fixed.In addition, the people involved in respond-ing to the emergency use the feedback to build their expertise.Resiliency is built over time by breaking things in order to findlatent problems and learn how to recover from the inevitable failuresthat every complex system harbors.This kind of testing reveals ev-erything from very tough problems to simple oversights.Quite a fewproblems originate in the product development process, so the learn-ing from a single event can move a long way back into the companyto change everything from design standards to testing approaches.Other companies with very large data centers Google, for examplehold similar events.Over time, these exercises have created the learn-ing, confidence, and resilience that allow companies to be comfortablewith massive complexity.There is no thought of perfection; it is simplyimpossible.There is no thought of declaring victory; learning throughfailure injection is a never-ending journey of constantly getting better.People grow in expertise and gain confidence in their ability to handlethe confusion of an outage, and the company s systems and processesbecome more hardened against catastrophe.The Paradox of PerfectionInitially, failure injection events were very hard to sell.Theprevailing wisdom was that data centers should not fail;they were measured by the percentage of uptime they couldguarantee and it was supposed to approach 100%.It was25.A discussion with Jesse Robbins, Kripa Krishnan, John Allspaw, and Tom Limon-celli, Resilience Engineering: Learning to Embrace Failure, Communications of theACM, November 2012. CHAPTER 2 ENERGIZED WORKERS 69thought that excellent data centers were the ones that were able toeliminate downtime; it seemed inappropriate for data centers to spendtime worrying about how fast they could recover from failures thatwere not supposed to happen.Intentionally causing a failure one thatcould have serious consequences is actually quite risky.And it is alsoexpensive; these exercises involve many people working around theclock for days.Companies that pursue resiliency through failure injection havecome to understand two basic facts:1.Perfection is impossible in a complex system at scale.No matterhow low the probability of a failure is, the number of transac-tions is so large that the only real question is when, not if, afailure will happen.2.It is a lot less expensive to develop the expertise to recoverquickly and safely from failure than it is to pretend that failurewill not happen.It can take a radical cultural shift for a company to accept the idea thatperfection is not the right goal; resilience is a much better goal in thecontext of complexity.And resiliency is built on, and builds, expertise.Questions to Ponder1.Appraisal and compensation practices seem to come in two fla-vors.One flavor supports a mindset that assumes talent is a fixedasset in a person.The other flavor supports a growth mindsetthat assumes talent can be significantly increased through hardwork and experience.Looking honestly and critically at the prac-tices in your company, which mindset do they reflect?2 [ Pobierz całość w formacie PDF ]
zanotowane.pl doc.pisz.pl pdf.pisz.pl agnieszka90.opx.pl
.23.Gary Klein, Streetlights and Shadows: Searching for the Keys to Adaptive Deci-sion Making (The MIT Press, 2009), p.13.24.Kahneman and Klein, Conditions for Intuitive Expertise. 68 THE LEAN MINDSETResiliencyWhat does developing expertise look like in practice? A good exam-ple can be found in the discussion Resilience Engineering: Learningto Embrace Failure. 25 In this article, Jesse Robbins discusses how hisfirefighter training led him to convince his company (Amazon.com)that the best way to keep their complex systems up and running wasnot to make them perfect, but to make them resilient.In order to dothis, he started GameDay events events that are triggered by inten-tionally pulling the plug on a data center (or a similar self-inflictedfailure).For a typical GameDay event, it may take dozens of peopletwo or three days to get the system back to normal.Once things areworking again, a detailed review occurs in which dependencies andprocess weaknesses exposed by the exercise are documented and mostof them are rapidly fixed.In addition, the people involved in respond-ing to the emergency use the feedback to build their expertise.Resiliency is built over time by breaking things in order to findlatent problems and learn how to recover from the inevitable failuresthat every complex system harbors.This kind of testing reveals ev-erything from very tough problems to simple oversights.Quite a fewproblems originate in the product development process, so the learn-ing from a single event can move a long way back into the companyto change everything from design standards to testing approaches.Other companies with very large data centers Google, for examplehold similar events.Over time, these exercises have created the learn-ing, confidence, and resilience that allow companies to be comfortablewith massive complexity.There is no thought of perfection; it is simplyimpossible.There is no thought of declaring victory; learning throughfailure injection is a never-ending journey of constantly getting better.People grow in expertise and gain confidence in their ability to handlethe confusion of an outage, and the company s systems and processesbecome more hardened against catastrophe.The Paradox of PerfectionInitially, failure injection events were very hard to sell.Theprevailing wisdom was that data centers should not fail;they were measured by the percentage of uptime they couldguarantee and it was supposed to approach 100%.It was25.A discussion with Jesse Robbins, Kripa Krishnan, John Allspaw, and Tom Limon-celli, Resilience Engineering: Learning to Embrace Failure, Communications of theACM, November 2012. CHAPTER 2 ENERGIZED WORKERS 69thought that excellent data centers were the ones that were able toeliminate downtime; it seemed inappropriate for data centers to spendtime worrying about how fast they could recover from failures thatwere not supposed to happen.Intentionally causing a failure one thatcould have serious consequences is actually quite risky.And it is alsoexpensive; these exercises involve many people working around theclock for days.Companies that pursue resiliency through failure injection havecome to understand two basic facts:1.Perfection is impossible in a complex system at scale.No matterhow low the probability of a failure is, the number of transac-tions is so large that the only real question is when, not if, afailure will happen.2.It is a lot less expensive to develop the expertise to recoverquickly and safely from failure than it is to pretend that failurewill not happen.It can take a radical cultural shift for a company to accept the idea thatperfection is not the right goal; resilience is a much better goal in thecontext of complexity.And resiliency is built on, and builds, expertise.Questions to Ponder1.Appraisal and compensation practices seem to come in two fla-vors.One flavor supports a mindset that assumes talent is a fixedasset in a person.The other flavor supports a growth mindsetthat assumes talent can be significantly increased through hardwork and experience.Looking honestly and critically at the prac-tices in your company, which mindset do they reflect?2 [ Pobierz całość w formacie PDF ]