1 00:00:07,500 --> 00:00:11,420 After watching this video, you will be able to explain the importance of measuring what 2 00:00:11,420 --> 00:00:15,980 you want to improve, identify the value of social and DevOps metrics, 3 00:00:15,980 --> 00:00:19,980 and recognize that DevOps changes your approach to problem resolution. 4 00:00:19,980 --> 00:00:22,189 How do you change a culture? 5 00:00:22,189 --> 00:00:27,390 This is from a paper by Steven Kerr from the Academy Management Journal back in 1975 on 6 00:00:27,390 --> 00:00:30,890 “The folly of rewarding for A, while hoping for B.” 7 00:00:30,890 --> 00:00:35,480 “Whether dealing with monkeys, rats, or human beings, it is hardly controversial to 8 00:00:35,480 --> 00:00:40,700 state that most organisms seek information concerning what activities are rewarded, and 9 00:00:40,700 --> 00:00:46,110 then seek to do (or at least pretend to do) those things, often to the virtual exclusion 10 00:00:46,110 --> 00:00:48,770 of activities not rewarded. 11 00:00:48,770 --> 00:00:52,420 The extent to which this occurs, of course, will depend on the perceived attractiveness 12 00:00:52,420 --> 00:00:57,200 of the reward offered, but neither operant nor expectancy theorists would quarrel with 13 00:00:57,200 --> 00:00:59,670 the essence of this notion.” 14 00:00:59,670 --> 00:01:05,570 You cannot measure for A and hope you get B. It’s not going to happen because you 15 00:01:05,570 --> 00:01:07,370 get what you measure. 16 00:01:07,370 --> 00:01:09,840 So, measure what matters. 17 00:01:09,840 --> 00:01:13,850 If you measure widget production, you will get lots of widgets. 18 00:01:13,850 --> 00:01:17,659 If you measure lines of code, you will get many lines of code. 19 00:01:17,659 --> 00:01:19,520 Semicolons all the way down. 20 00:01:19,520 --> 00:01:23,890 We used to measure lines of code by the thousands of lines of code (KLOC). 21 00:01:23,890 --> 00:01:24,890 Is it good code? 22 00:01:24,890 --> 00:01:27,280 I do not know; I do not care. 23 00:01:27,280 --> 00:01:31,590 I am measured by how many lines of code I write, so I am going to write verbose code. 24 00:01:31,590 --> 00:01:34,380 Remember, you get what you measure. 25 00:01:34,380 --> 00:01:39,200 If you measure people by ranking them against each other, you will get antisocial behavior. 26 00:01:39,200 --> 00:01:41,790 A lot of companies do this—they rank people. 27 00:01:41,790 --> 00:01:43,440 So, let me get this straight. 28 00:01:43,440 --> 00:01:47,100 You want me to help my peer, but you are going to rank us against each other? 29 00:01:47,100 --> 00:01:52,189 So, I help them so that they get a better ranking than me, then they get the pay raise 30 00:01:52,189 --> 00:01:53,890 or the promotion that I want? 31 00:01:53,890 --> 00:01:55,690 Yeah, how does that work? 32 00:01:55,690 --> 00:02:00,189 Companies do this all the time, and their management doesn’t seem to get it. 33 00:02:00,189 --> 00:02:01,619 You get what you measure. 34 00:02:01,619 --> 00:02:05,930 If you want people to be social, then you must measure them on being social. 35 00:02:05,930 --> 00:02:11,420 You should measure developers by their social interactions and sharing of code and knowledge. 36 00:02:11,420 --> 00:02:14,760 That is how you get social coders. 37 00:02:14,760 --> 00:02:18,950 Here are two metrics that I like to use to measure social coders. 38 00:02:18,950 --> 00:02:21,700 Who is leveraging the code you are building? 39 00:02:21,700 --> 00:02:25,290 Are you building code that the rest of the company or open-source community find valuable 40 00:02:25,290 --> 00:02:28,120 enough to reuse in their solutions? 41 00:02:28,120 --> 00:02:29,760 But that’s only half of the puzzle. 42 00:02:29,760 --> 00:02:34,540 That will get people thinking about how can they make their code valuable to others. 43 00:02:34,540 --> 00:02:36,540 But you need the second metric. 44 00:02:36,540 --> 00:02:38,480 Whose code are you leveraging? 45 00:02:38,480 --> 00:02:43,540 Are you wasting all of my development dollars reinventing wheels or are you leveraging the 46 00:02:43,540 --> 00:02:48,760 existing wheels and only building the bespoke parts that don’t exist? 47 00:02:48,760 --> 00:02:54,269 Both of these metrics are needed because they incentivize both sides to share their code 48 00:02:54,269 --> 00:02:58,690 and reuse each other’s code because… you get what you measure. 49 00:02:58,690 --> 00:03:02,459 You need a social metric to get social behavior. 50 00:03:02,459 --> 00:03:05,790 DevOps is all about continuous improvement. 51 00:03:05,790 --> 00:03:09,349 That means that you must know where you are coming from in order to measure if you are 52 00:03:09,349 --> 00:03:11,480 going in the right direction. 53 00:03:11,480 --> 00:03:13,400 It starts with taking a baseline. 54 00:03:13,400 --> 00:03:14,800 What do you want to improve? 55 00:03:14,800 --> 00:03:20,750 Maybe it currently requires six team members 10 hours to deploy a release of your product. 56 00:03:20,750 --> 00:03:22,950 Maybe it costs X dollars for every release. 57 00:03:22,950 --> 00:03:26,470 Whatever that measurement is, that’s the baseline. 58 00:03:26,470 --> 00:03:28,550 Then you need to create a goal. 59 00:03:28,550 --> 00:03:33,700 Metric goals allow you to reason about these numbers and judge the success of your progress. 60 00:03:33,700 --> 00:03:36,659 Perhaps you want to reduce deployment time from 10 hours to 2 hours. 61 00:03:36,659 --> 00:03:39,069 That’s a very achievable goal. 62 00:03:39,069 --> 00:03:43,599 Your baseline determines that it takes 10 hours, so you aim for a threshold of fivefold 63 00:03:43,599 --> 00:03:45,560 improvement to 2 hours. 64 00:03:45,560 --> 00:03:50,090 Or maybe you want to go from needing 6 team members to one team member. 65 00:03:50,090 --> 00:03:52,840 Or maybe you want to reduce the number of defects found in production. 66 00:03:52,840 --> 00:03:58,480 Whatever it is, choose a goal and work on one at a time. 67 00:03:58,480 --> 00:04:00,650 Then you judge the success of your process. 68 00:04:00,650 --> 00:04:03,129 Did you go from 10 hours to 2 hours? 69 00:04:03,129 --> 00:04:04,129 No? 70 00:04:04,129 --> 00:04:06,849 Try something different until you achieve this measurement. 71 00:04:06,849 --> 00:04:08,909 Then you start working on the next goal. 72 00:04:08,909 --> 00:04:11,200 DevOps changes the objective. 73 00:04:11,200 --> 00:04:15,730 We used to measure mean time to failure where you tried to make sure the server never went 74 00:04:15,730 --> 00:04:17,090 down. 75 00:04:17,090 --> 00:04:19,120 That is old-school thinking. 76 00:04:19,120 --> 00:04:22,289 We need to go from mean time to failure to mean time to recovery. 77 00:04:22,289 --> 00:04:24,470 You anticipate that the server will go down. 78 00:04:24,470 --> 00:04:27,000 Just be sure you can recover quickly. 79 00:04:27,000 --> 00:04:32,050 If your applications are built as a collection of microservices deployed in containers as 80 00:04:32,050 --> 00:04:36,310 multiple instances, you can recover quickly by bringing up a new container. 81 00:04:36,310 --> 00:04:40,740 And your customer might never even know that the service went down because there was no 82 00:04:40,740 --> 00:04:43,259 single point of failure. 83 00:04:43,259 --> 00:04:47,840 Containers may be continuously going up and down, but if you can recover quickly by spinning 84 00:04:47,840 --> 00:04:50,120 up a new container, who cares it went down? 85 00:04:50,120 --> 00:04:51,400 The customer never noticed. 86 00:04:51,400 --> 00:04:56,199 It’s a whole new way of thinking about availability. 87 00:04:56,199 --> 00:05:01,410 In this video, you learned that you should measure and reward what you want to improve. 88 00:05:01,410 --> 00:05:06,169 Measuring social metrics leads to improved socialization and measuring DevOps metrics 89 00:05:06,169 --> 00:05:09,340 allows you to see progression toward goals. 90 00:05:09,340 --> 00:05:15,862 DevOps changes the objective of problem resolution from failure prevention to failure recovery.