Top Banner
software simulated memories machines William louth
54
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Google TechTalk Stockholm 13th May 2015

software simulated

memories machines

William louth

Page 2: Google TechTalk Stockholm 13th May 2015

a new hope disillusionment

jinspired BV autoletics

personal MEMORIES

INSPIRED LTD Borland

at&T HP OpenView

APM TRANSACTION PATH TRACING

SOLUTIONS

DISTANT past 2008

ABC/M ACTIVITY BASED

COSTING & METERING SOFTWARE SOLUTION

CARS COST AWARE

RUNTIME & SERVICES

2009 2010

QOS 4 APPS APPLICATION LEVEL QUALITY OF SERVICE

stenos SOFTWARE METERING RECORD & PLAYBACK

2011

I.AM INTELLIGENT ACTIVITY

METERING

UDR BILLING USAGE DATA

BTM Supervisory control &

BEHAVIORAL self-REFLECTION

2012

SIMZ ONLINE SOFTWARE

MACHINE MIRRORING

ACE ADAPTIVE CONTROL

IN EXECUTION “flow valves”

2013

signals self adaptive

software runtimes

2015

charge back has its costs

show back is a no SHOW in “IT”

quality is elusive when it must be defined by “IT”

to recollect in order to learn

requires attention

awareness without action is

worthless

“IT” is delusional it thinks it is in

control when it is not

“IT” holds data over action

“IT” hates self-adaptation

more than change itself

what i learnt the hard way

Page 3: Google TechTalk Stockholm 13th May 2015

machines machines

Page 4: Google TechTalk Stockholm 13th May 2015

Mirrored simulation

mechanical sympathy

algorithms & structurescode

optimization

execution heuristics

software adaptation

system dynamics

execution caches

the underworld Elysium

software memorization

execution parallelism

dataset training

online learning

embodiment & augmentation

emergent behavior

pipelining & prediction

software execution

model

system execution

model

corruptive influence

collective intelligence

fast limitless

Page 5: Google TechTalk Stockholm 13th May 2015

context context

Page 6: Google TechTalk Stockholm 13th May 2015

Cognizant Computing

Page 7: Google TechTalk Stockholm 13th May 2015

Cognizant machines

Page 8: Google TechTalk Stockholm 13th May 2015

cause effect

microservices

mobile & IoT

continuous delivery

many more small actions

many more small changes

many more small systems

consolidation

recollection

observation

mirror and simulate

Probe and measure

record and playback

“immutable” infrastructuremany more

small images

differentiationbehavioral delta analysis

Page 9: Google TechTalk Stockholm 13th May 2015

seeing seeing

Page 10: Google TechTalk Stockholm 13th May 2015

service

command coordinate

service

service call

service

event notify

action

service

servicepercept

👀

✉✉

Page 11: Google TechTalk Stockholm 13th May 2015

threads simulated

a

b

c

d

a > B B > C C > D

methods

“a”

“c”

“d”

probes

“a”

“c”

“d”

probes

executed

call stack

frames

simulatedmetered

Page 12: Google TechTalk Stockholm 13th May 2015

machines simulated

threads threads

Page 13: Google TechTalk Stockholm 13th May 2015

machine mirroring

execution simulationprojection

Page 14: Google TechTalk Stockholm 13th May 2015

☯event

action percept

call

Perception developed from action to serve action

shaped by learning, memory, expectation, and attention

Page 15: Google TechTalk Stockholm 13th May 2015

stack sequence

a

b

c

d

top

call depth d c b a

begin begin begin begin first

event stream

Page 16: Google TechTalk Stockholm 13th May 2015

pop add

a

bcall

depth

C Dend end

event stream

top

first

Page 17: Google TechTalk Stockholm 13th May 2015

memories memories

Page 18: Google TechTalk Stockholm 13th May 2015

life without memory would be close to impossible

software has memory but no memories

Page 19: Google TechTalk Stockholm 13th May 2015

human memory

sensory memory

short-term memory

long-term memory

explicit memory

implicit memory

declarative memory

procedural memory

episodic memory

semantic memory

events experiences

facts concepts

skills tasks

memory models

Page 20: Google TechTalk Stockholm 13th May 2015

remembering eventually

Page 21: Google TechTalk Stockholm 13th May 2015

Deckard “Memories!

You're talking about memories!”

UNICORN DREAMS

blade runner

Tyrell “If we gift them with a past,

we create a cushion or a pillow for their emotions,

and consequently, we can control them better.”

Page 22: Google TechTalk Stockholm 13th May 2015

time time

Page 23: Google TechTalk Stockholm 13th May 2015

time’s arrow is straight with one exception being remembrance

the reality of the mind is at least as important as the physical reality

time less

Page 24: Google TechTalk Stockholm 13th May 2015

Do not try and bend the spoon—that’s impossible.

time now

Instead, only try to realize the truth. There is no spoon.

Page 25: Google TechTalk Stockholm 13th May 2015

software mirroring

demo

simulated playback

Page 26: Google TechTalk Stockholm 13th May 2015

mind mind

Page 27: Google TechTalk Stockholm 13th May 2015

cognition is strongly influenced by aspects of the

body beyond the brain

The function of the mind is to guide action

Page 28: Google TechTalk Stockholm 13th May 2015

every good regulator of a system must be a

model of that system

All doing is knowing and all knowing is doing

Page 29: Google TechTalk Stockholm 13th May 2015

We form predictions…and explanations of someone by putting……ourselves in the shoes

of another person…..and simulating them

Page 30: Google TechTalk Stockholm 13th May 2015

motion motion

Page 31: Google TechTalk Stockholm 13th May 2015

Act

agent

agency

scenepurpose

attitude

LIFE DRAMA

Page 32: Google TechTalk Stockholm 13th May 2015

ACT What was done?

SCENE Where was it done?

Agent Who did it?

Purpose Why did it happen?

Agency how was it done?

DRAMATISM PROFILINGWHAT CRIME (ACT) COMMITTED

WHERE SCENE WHERE IT HAPPENED

WHEN TIME ACT TOOK PLACE

WHO AGENT (ACTOR)

WHY MOTIVE (INTENT)

HOW PERFORMANCE (ACTION)

Page 33: Google TechTalk Stockholm 13th May 2015

activity metering

Activity Resource

Device Probe Meter

Develop Code Counter

Design Behavior Usage

Data Group Metering

Page 34: Google TechTalk Stockholm 13th May 2015

Context+Thread

Probe =Code

MeteringMeter+Resource

actoractivity

begin event end event

call

act(ion)

execution metering

Page 35: Google TechTalk Stockholm 13th May 2015

Actor Activity

Context

Coordinate

Event

Environment

Resource

Page 36: Google TechTalk Stockholm 13th May 2015

motion picture

Page 37: Google TechTalk Stockholm 13th May 2015

monitoring monitoring

Page 38: Google TechTalk Stockholm 13th May 2015

Application #1

The view the service provider has of the infrastructure and applications.

Application

The view a PaaS vendor or managed service provider offers to customers.

Application

The view a customer needs of applications deployed to PaaS.

Infrastructure Infrastructure

Application #2

Application #3

cloudy visibility

Page 39: Google TechTalk Stockholm 13th May 2015

010101010101010101

01010101010

010101010101010101

010101010101

010101010101010

010101010

01010101010

0101010101010

010101010101010101

0101010101010

01010101

010101010

010101010101010101

01010101010

010101010101010101

010101010101

010101010101010

010101010

0101010101010

010101010101010101

0101010101010

01010101

010101010

010101010101010101

01010101010

010101010101010101

010101010101

010101010101010

010101010

01010101010

0101010101010

010101010101010101

0101010101010

01010101

010101010

010101010101010101

01010101010

Threads

0101

01010

0101010

0101

Simz

trinity neo morpheus Threads

0101

01010

0101010

0101

Simz

Threads

0101

01010

0101010

0101

Simz

users apps

mirrored machines

Page 40: Google TechTalk Stockholm 13th May 2015

Infrastructure

Application #1

Customer Dev(Ops) & Test

Teams

Mirroring & Simulation

Simz Service

Single Application Monitoring

Plane

Metering Feed

UnidirectionalApplication #2

Application #3

mirror isolate

Page 41: Google TechTalk Stockholm 13th May 2015

Infrastructure

Application #1

Security & Auditing

Mirroring & Simulation

Simz Service

Simulated Application

Metering Feed

UnidirectionalApplication #2

Application #3

Business Analytics & Integrations

Service Level Management

API

API

mirror augment

Page 42: Google TechTalk Stockholm 13th May 2015

collective intelligence

post-execution augmentation

demo

Page 43: Google TechTalk Stockholm 13th May 2015

ping pong

simz playback

satoris agent

simz client

Precept

game

Precept

set state

metered activity

replicate action

simz server

satoris agent

simz playback

metering engine

simz server

Precept

game

Precept

set state

metered activity

simz client

simulate action

simulate action

replicate action

metering feed

ping

Pong

action > pingprecept < pong

pong < actionping > precept

Page 44: Google TechTalk Stockholm 13th May 2015

PING PONG

SEEING MINDS

demo

Page 45: Google TechTalk Stockholm 13th May 2015

POSSIBILITIES POSSIBILITIES

Page 46: Google TechTalk Stockholm 13th May 2015

mini-me computing

real application

process

heap

IOthreads

code

probes

stacks

threads

probes

stacks

mirrored application

process

isolation

metered behavio

r

diagnostics supervision ————————

signaling symbiosis

episodic memory

crash report

Page 47: Google TechTalk Stockholm 13th May 2015

anytime computing

online

application

mirroring

offline

recording playback

integrations

integrations

service monitoring security & auditing

workflow integration

change detection software analytics post augmentation

time

simulation

simulation

Page 48: Google TechTalk Stockholm 13th May 2015

in-flight simulator

playback

OK trouble trouble

crash

timelinex x

“total recall” for devops deviation deviation

Page 49: Google TechTalk Stockholm 13th May 2015

polyglot universe

mirrored application

process

isolation

metered behavio

r

episodic memory

scala

java

ruby

javascript

java plugins ————— probes

open api

Page 50: Google TechTalk Stockholm 13th May 2015

FINALLY FINALLY

Page 51: Google TechTalk Stockholm 13th May 2015

SCRIPT RECALL

simulates & playbacks software behavior

simulates & playbacks

request data

data does not scale

✉✉✉✉

540 million

events/sec

coupled to code

& version

there is no code

only action

what was actioned

what was asked

can dreamzzzzzzzz

immutablerollback

Page 52: Google TechTalk Stockholm 13th May 2015

reading reading

Page 53: Google TechTalk Stockholm 13th May 2015
Page 54: Google TechTalk Stockholm 13th May 2015

@autoleticsfollow