Critical Path Analysis - Speaker Deck

Slide 1

Slide 1 text

@rakyll

Slide 2

Slide 2 text

@rakyll

Slide 3

Slide 3 text

@rakyll We use facilities that are not reserved for our flight but are essential in making it happen.

Slide 4

Slide 4 text

@rakyll We use facilities that are not reserved for our flight but are essential in making it happen.

Slide 5

Slide 5 text

@rakyll

Slide 6

Slide 6 text

@rakyll Systems? who understands them?

Slide 7

Slide 7 text

@rakyll Jaana B. Dogan jbd@google.com

Slide 8

Slide 8 text

@rakyll

Slide 9

Slide 9 text

@rakyll Growing...

Slide 10

Slide 10 text

@rakyll Tech stack gets fragmented, monoliths are gone...

Slide 11

Slide 11 text

@rakyll one becomes many failure in isolation escalation in failure?

Slide 12

Slide 12 text

@rakyll and it may get larger...

Slide 13

Slide 13 text

@rakyll good guy jeff

Slide 14

Slide 14 text

@rakyll

Slide 15

Slide 15 text

@rakyll

Slide 16

Slide 16 text

@rakyll code search

Slide 17

Slide 17 text

@rakyll go_library( name = "logs", srcs = ["logs.go"], visibility = ["//visibility:public"], deps = [ …. ], ) References (641 occurrences) - //source/ads/monitoring/BUILD - //source/ads/analysis/BUILD - //source/ads/mobile/BUILD ...

Slide 18

Slide 18 text

@rakyll frontend server authentication users images memcache blobservice memcache memcache (metadata) (disks) load balancer

Slide 19

Slide 19 text

@rakyll frontend server authentication users images memcache blobservice memcache memcache (metadata) (disks) load balancer critical path

Slide 20

Slide 20 text

@rakyll frontend server authentication users images memcache blobservice memcache memcache (metadata) (disks) load balancer critical path

Slide 21

Slide 21 text

@rakyll frontend server authentication users images memcache blobservice memcache memcache (metadata) (disks) load balancer critical path

Slide 22

Slide 22 text

@rakyll CPA (critical path analysis)

Slide 23

Slide 23 text

@rakyll discover the critical paths make them reliable and fast make them debuggable

Slide 24

Slide 24 text

@rakyll How do we get there? (tracing, events, logs)

Slide 25

Slide 25 text

@rakyll Why? Why? Why? Why? Why?

Slide 26

Slide 26 text

@rakyll GET /timeline edge-lb sched api-server auth.Auth cache.Get mysql.Query user.Profile cache.Get mysql.Query images.Filter blobstore.Get

Slide 27

Slide 27 text

@rakyll bare metal kernel network stack cloud stack libraries frameworks your code

Slide 28

Slide 28 text

@rakyll GET /timeline edge-lb sched api-server auth.Auth cache.Get mysql.Query user.Profile cache.Get mysql.Query images.Filter blobstore.Get not my fault

Slide 29

Slide 29 text

@rakyll GET /timeline auth.Auth cache.Get mysql.Query user.Profile cache.Get mysql.Query images.Filter blobstore.Get cache.Get mysql.Query blob.Get where is the source code?

Slide 30

Slide 30 text

@rakyll GET /timeline auth.Auth cache.Get mysql.Query user.Profile cache.Get mysql.Query images.Filter blobstore.Get cache.Get mysql.Query blob.Get who to call?

Slide 31

Slide 31 text

@rakyll GET /timeline auth.Auth cache.Get mysql.Query user.Profile cache.Get mysql.Query images.Filter blobstore.Get cache.Get mysql.Query blob.Get give me the logs, runtime events, profiles...

Slide 32

Slide 32 text

@rakyll Challenges...

Slide 33

Slide 33 text

@rakyll This is an organizational problem. CHALLENGE #1:

Slide 34

Slide 34 text

@rakyll Engineers don’t know where to start. CHALLENGE #2:

Slide 35

Slide 35 text

@rakyll Infra is still a blackbox. CHALLENGE #3:

Slide 36

Slide 36 text

@rakyll Instrumentation is expensive. CHALLENGE #4:

Slide 37

Slide 37 text

@rakyll Dynamic capabilities are underestimated. CHALLENGE #5:

Slide 38

Slide 38 text

@rakyll

Slide 39

Slide 39 text

@rakyll A tool to close knowledge gaps (which we don’t talk about)

Slide 40

Slide 40 text

@rakyll Fin jbd@google.com