Slide 1

Slide 1 text

CORRELATION: THE NEXT FRONTIER monitorama 2013 / boston / @aq

Slide 2

Slide 2 text

CTO of

Slide 3

Slide 3 text

Chief Taco Officer CTO of

Slide 4

Slide 4 text

No content

Slide 5

Slide 5 text

No content

Slide 6

Slide 6 text

A litte bit of @aq

Slide 7

Slide 7 text

A litte bit of @aq • Expert Eater

Slide 8

Slide 8 text

A litte bit of @aq • Expert Eater • Experienced Ruby and JS Developer

Slide 9

Slide 9 text

A litte bit of @aq • Expert Eater • Experienced Ruby and JS Developer • Growing Student of Operations

Slide 10

Slide 10 text

A litte bit of @aq • Expert Eater • Experienced Ruby and JS Developer • Growing Student of Operations • Beginner Distributed Systems Maintainer

Slide 11

Slide 11 text

Give each other dap. WE did it

Slide 12

Slide 12 text

OPS DONE. WE have DATA Y’ALL

Slide 13

Slide 13 text

No content

Slide 14

Slide 14 text

Uhhh SO NO MORE TEARS, right?

Slide 15

Slide 15 text

No content

Slide 16

Slide 16 text

THE LEVELS OF MONITORING NIRVANA

Slide 17

Slide 17 text

No content

Slide 18

Slide 18 text

PURE DATA

Slide 19

Slide 19 text

PURE DATA BASIC INFERENCES AND CORRELATIONS

Slide 20

Slide 20 text

PURE DATA BASIC INFERENCES AND CORRELATIONS THE FUCKING MATRIX

Slide 21

Slide 21 text

PURE DATA BASIC INFERENCES AND CORRELATIONS THE FUCKING MATRIX

Slide 22

Slide 22 text

PURE DATA BASIC INFERENCES AND CORRELATIONS THE FUCKING MATRIX whoa

Slide 23

Slide 23 text

No content

Slide 24

Slide 24 text

PURE DATA

Slide 25

Slide 25 text

PURE DATA BASIC INFERENCES AND CORRELATIONS

Slide 26

Slide 26 text

PURE DATA BASIC INFERENCES AND CORRELATIONS PREDICTIVE AND DIRECT RELATIONSHIPS

Slide 27

Slide 27 text

Aligning the data. CORRELATION

Slide 28

Slide 28 text

Except when it does. CORRELATION DOES NOT IMPLY CAUSATION

Slide 29

Slide 29 text

The marshmallow test

Slide 30

Slide 30 text

And let us get back to Shaving Yaks. correlation can narrow our work

Slide 31

Slide 31 text

Step back, HE’S DOING MATH! MATHEMATICAL CORRELATION

Slide 32

Slide 32 text

Say that 5 times fast. PEARSON Product moment correlation coefficient

Slide 33

Slide 33 text

0 1.5 3 4.5 6 0 225 450 675 900 CPU vs Response Time

Slide 34

Slide 34 text

0 1.5 3 4.5 6 0 225 450 675 900 CPU vs Response Time

Slide 35

Slide 35 text

No content

Slide 36

Slide 36 text

1 data = [ 2 [100, 0.7], 3 [125, 0.5], 4 [150, 1], 5 [300, 2.1], 6 [500, 3.4], 7 [900, 6] 8 ] 9 10 x, y = data.transpose 11 n = data.size 12 x_mean = x.reduce(:+) / n 13 y_mean = y.reduce(:+) / n 14 x_stddev = Math.sqrt(x.inject {|sum, i| sum + (i - x_mean)**2 } / (n - 1).to_f) 15 y_stddev = Math.sqrt(y.inject {|sum, i| sum + (i - y_mean)**2 } / (n - 1).to_f) 16 z_x = x.collect {|i| (i - x_mean) / x_stddev } 17 z_y = y.collect {|i| (i - y_mean) / y_stddev } 18 pearsons = z_x.zip(z_y).collect {|x| x[0] * x[1] }.reduce(:+) / n 19 # => 0.9265763490538744

Slide 37

Slide 37 text

PEarson

Slide 38

Slide 38 text

PEarson • Close to absolute 1 = probably correlated samples

Slide 39

Slide 39 text

PEarson • Close to absolute 1 = probably correlated samples • Could be applied to moving averages?

Slide 40

Slide 40 text

PEarson • Close to absolute 1 = probably correlated samples • Could be applied to moving averages? • Could we pull it into a graphite function? (Hackathon anyone?)

Slide 41

Slide 41 text

LIMITS OF Mathematical correlation

Slide 42

Slide 42 text

LIMITS OF Mathematical correlation • Requires known inputs and assumptions

Slide 43

Slide 43 text

LIMITS OF Mathematical correlation • Requires known inputs and assumptions • Suggestion of correlation, not proof

Slide 44

Slide 44 text

LIMITS OF Mathematical correlation • Requires known inputs and assumptions • Suggestion of correlation, not proof • Needs a large amount of knowledge of the data set to make decisions

Slide 45

Slide 45 text

I can see it! TIME BASED VISUAL CORRELATION

Slide 46

Slide 46 text

A Graphite Story

Slide 47

Slide 47 text

LIMITS OF VISUAL CORRELATION

Slide 48

Slide 48 text

LIMITS OF VISUAL CORRELATION • Takes a good eye

Slide 49

Slide 49 text

LIMITS OF VISUAL CORRELATION • Takes a good eye • Hard to see the signal through the noise

Slide 50

Slide 50 text

LIMITS OF VISUAL CORRELATION • Takes a good eye • Hard to see the signal through the noise • Doesn’t really account for domino events

Slide 51

Slide 51 text

LIMITS OF VISUAL CORRELATION • Takes a good eye • Hard to see the signal through the noise • Doesn’t really account for domino events • Good for trends but not as much for events

Slide 52

Slide 52 text

There’s a disturbance in the force. EMOTIONAL CORRELATION

Slide 53

Slide 53 text

RASHoMONING

Slide 54

Slide 54 text

Each person uses their unique knowledge of the situation to point out unique data points. RASHoMONING

Slide 55

Slide 55 text

LIMITS OF EMOTIONAL correlation

Slide 56

Slide 56 text

LIMITS OF EMOTIONAL correlation • Provides a trail not an answer

Slide 57

Slide 57 text

LIMITS OF EMOTIONAL correlation • Provides a trail not an answer • Depends on having a team of people

Slide 58

Slide 58 text

LIMITS OF EMOTIONAL correlation • Provides a trail not an answer • Depends on having a team of people • Many ideas, needs a “judge”

Slide 59

Slide 59 text

LIMITS OF EMOTIONAL correlation • Provides a trail not an answer • Depends on having a team of people • Many ideas, needs a “judge” • HUMANS (Hence Rashomoning)

Slide 60

Slide 60 text

I’m sold, show me how!

Slide 61

Slide 61 text

But I have some ideas. I don’t know exactly

Slide 62

Slide 62 text

TRYING TO MAKE THE DATA MORE VISIBLE

Slide 63

Slide 63 text

A conflagration of data HOTPOT

Slide 64

Slide 64 text

No content

Slide 65

Slide 65 text

Hotpot = Chef, Sensu, Graphite, (Logstash) Simply align disparate sources of data TO VISUALLY CORRELATE

Slide 66

Slide 66 text

create relationships for alerts and notifications

Slide 67

Slide 67 text

No content

Slide 68

Slide 68 text

Math to filter out noise. USe PEARSONS to pull out potentially related data

Slide 69

Slide 69 text

Have the ability to easily divide datasets by “cohorts” cohort analysis for processes/nodes

Slide 70

Slide 70 text

“Node notes”. Document everything. Treat personal and institutional knowledge as data

Slide 71

Slide 71 text

By making more data available to everyone. Make emotional correlation less EMO

Slide 72

Slide 72 text

So if that’s just level 2, what’s level 3?

Slide 73

Slide 73 text

No content

Slide 74

Slide 74 text

TAKE the correlations and let the machine turn them into decisions

Slide 75

Slide 75 text

Lets figure out level 2 first. NOT YET.

Slide 76

Slide 76 text

github.com/quirkey github.com/paperlesspost twitter.com/aq twitter.com/paperlessdev quirkey.com paperlesspost.com THANKS!