openshift for ai and ml -...

OpenShift for AI and MLWilliam Benton • @willb • willb@redhat.com

@willb

OpenShift is enterprise Kubernetes with a great developer experience.

@willb

resource managementfor apps and compute

@willb

efficient isolation, secure by default

@willb

workflows to accelerate discovery

What do machine learning workflows look like?

@willb

f( ) = 0.67 0.57 0.84 0.08 0.42 0.01

@willb

data collection and cleaning

codifying problem and metrics

@willb

feature engineering

model training and tuning

@willb

feature engineering

@willb

feature engineering

model validation

@willb

feature engineering

model validation

@willb

model validation

model deployment

monitoring and validation

@willb

model validation

model deployment

@willb

feature engineering

model validation

model deployment

@willb

feature engineering

model validation

model deployment

defining types and interfaces prototyping

@willb

feature engineering

model validation

model deployment

unit, behavioral, and integration testing

formal verification

@willb

feature engineering

model validation

model deployment

deployment monitoring

@willb

Why is OpenShift great for machine learning workflows?

@willb

% pip install numpy

@willb

pip install numpy

/usr/bin/pip executable

arguments

virtual memory

file handles

/root filesystem

environment LANG=en_US USER=willb ...

process table

network routes

@willb

pip install numpy

arguments

virtual memory

file handles

/root filesystem

process table

network routes

@willb

pip install numpy

arguments

virtual memory

file handles

/root filesystem

process table

network routes

@willb

pip install numpy

arguments

virtual memory

file handles

/root filesystem

process table

network routes

@willb

pip install numpy

arguments

virtual memory

file handles

/root filesystem

process table

network routes

@willb

pip install numpy

arguments

virtual memory

file handles

/var/lib/envs/mainroot filesystem

process table

network routes

@willb

pip install numpy

arguments

virtual memory

file handles

/var/lib/envs/mainroot filesystem

process table

network routes

SPEED LIMIT

@willb

Immutable images

base image

configuration and installation recipes

user application code

979229b9

33721112 e8cae4f6 2bb6ab16 a8296f7e

a6afd91e 6b8cad3e

@willb

Stateless microservices

@willb

Declarative app configuration

@willb

Integration and deployment

@willb

OK!base image

application code

@willb

OK!base image

application code

@willb

base image

application code

What OpenShift offers working data scientists

@willb

workflows to accelerate discovery

@willb

0 0 0 1 1 0 1 0 1 0

0 0 1 0 0 0 1 1 0 0

1 0 1 1 0 1 0 0 0 0

0 0 0 0 0 0 1 1 0 1

0 1 0 0 1 0 0 1 0 0

1 0 0 0 0 1 0 1 1 0

0 0 1 0 1 0 1 0 0 0

0 1 0 0 0 1 0 0 1 1

0 0 0 0 1 0 0 1 0 1

1 1 0 0 0 0 0 0 0 1

0.13 0.13

0.06 0.07

0.07 0.06

0.02 0.08

0.17 0.11

0.11 0.09

0.04 0.18

0.13 0.04

0.13 0.21

0.14 0.03

@willb

Self-service environments

more storagesensitive data

more CPUsbetter GPUs

@willb

No friction: mybinder.org

@willb

More flexible: source-to-image

@willb

More flexible: source-to-image

https://github.com/openshift/source-to-image

builder image application image

@willb

distribution of input data?

distribution of predictions?

distribution of acyclic paths taken through scoring code?

(joint)

@willb

Where from here?

@willb

Intelligent applications continuously collect and learn from data to support essential functionality

@willb

federate

trainmodels

events

databases

file, object storage

management

web and mobile

reporting

developer UItransform

transform

archive

@willb

federate

trainmodels

events

databases

management

web and mobile

reporting

transform

archive

@willbdata scientists

application developersdata engineers

federate

trainmodels

events

databases

management

web and mobile

reporting

transform

archive

application developersdata engineers

federate

trainmodels

events

databases

management

web and mobile

reporting

transform

archive

machine learning engineers

@willb

data engineers

federate

events

databases

transform

archive

federate

trainmodels

transform

@willb

application developers

federate

trainmodels management

web and mobile

reporting

transform

archive

@willb

Multitenant compute clustersCluster scheduler

Shared FS / object store

Compute node

Resource manager

app 1 app 2

app 4app 3 Databases

@willb

Multitenant compute clustersCluster scheduler

Shared FS / object store

Compute node

Resource manager

app 1 app 2

app 4app 3 Databases

@willb

One cluster per applicationOpenShift

Shared FS / object storeapp 1 app 2

app 5app 4

app 6 Databases

@willb

One cluster per applicationOpenShift

Shared FS / object storeapp 1 app 2

app 5app 4

app 4 Databases

@willb

radanalytics.io

@willb

opendatahub.io

@willb

Kubeflow

@willb

Other technologies

x · y = x1 · y1 + · · ·+ xn · yn<latexit sha1_base64="tjPm2BYBzcpGEXsTQhjwuSSRkxI=">AAACLXicbZBdS8MwFIbT+TXn19RLb4JDEITRiqA3wlAvvJzgPmAtJU3TLSxNS5LKStkf8sa/IoIXE/HWv2G6VdHNA4GH9z0nyXm9mFGpTHNilJaWV1bXyuuVjc2t7Z3q7l5bRonApIUjFomuhyRhlJOWooqRbiwICj1GOt7wOvc7D0RIGvF7lcbECVGf04BipLTkVm/sEKmBF2SjMbSxHyn4LaRjeAlHrlXIqaaTGUsNI5f/GPqamlk3pwUXwSqgBopqutUX249wEhKuMENS9iwzVk6GhKKYkXHFTiSJER6iPulp5Cgk0smm247hkVZ8GERCH67gVP09kaFQyjT0dGe+ipz3cvE/r5eo4MLJKI8TRTiePRQkDKoI5tFBnwqCFUs1ICyo/ivEAyQQVjrgig7Bml95EdqndcusW3dntcZVEUcZHIBDcAwscA4a4BY0QQtg8AiewQS8GU/Gq/FufMxaS0Yxsw/+lPH5BZIxpwo=</latexit><latexit sha1_base64="tjPm2BYBzcpGEXsTQhjwuSSRkxI=">AAACLXicbZBdS8MwFIbT+TXn19RLb4JDEITRiqA3wlAvvJzgPmAtJU3TLSxNS5LKStkf8sa/IoIXE/HWv2G6VdHNA4GH9z0nyXm9mFGpTHNilJaWV1bXyuuVjc2t7Z3q7l5bRonApIUjFomuhyRhlJOWooqRbiwICj1GOt7wOvc7D0RIGvF7lcbECVGf04BipLTkVm/sEKmBF2SjMbSxHyn4LaRjeAlHrlXIqaaTGUsNI5f/GPqamlk3pwUXwSqgBopqutUX249wEhKuMENS9iwzVk6GhKKYkXHFTiSJER6iPulp5Cgk0smm247hkVZ8GERCH67gVP09kaFQyjT0dGe+ipz3cvE/r5eo4MLJKI8TRTiePRQkDKoI5tFBnwqCFUs1ICyo/ivEAyQQVjrgig7Bml95EdqndcusW3dntcZVEUcZHIBDcAwscA4a4BY0QQtg8AiewQS8GU/Gq/FufMxaS0Yxsw/+lPH5BZIxpwo=</latexit><latexit sha1_base64="tjPm2BYBzcpGEXsTQhjwuSSRkxI=">AAACLXicbZBdS8MwFIbT+TXn19RLb4JDEITRiqA3wlAvvJzgPmAtJU3TLSxNS5LKStkf8sa/IoIXE/HWv2G6VdHNA4GH9z0nyXm9mFGpTHNilJaWV1bXyuuVjc2t7Z3q7l5bRonApIUjFomuhyRhlJOWooqRbiwICj1GOt7wOvc7D0RIGvF7lcbECVGf04BipLTkVm/sEKmBF2SjMbSxHyn4LaRjeAlHrlXIqaaTGUsNI5f/GPqamlk3pwUXwSqgBopqutUX249wEhKuMENS9iwzVk6GhKKYkXHFTiSJER6iPulp5Cgk0smm247hkVZ8GERCH67gVP09kaFQyjT0dGe+ipz3cvE/r5eo4MLJKI8TRTiePRQkDKoI5tFBnwqCFUs1ICyo/ivEAyQQVjrgig7Bml95EdqndcusW3dntcZVEUcZHIBDcAwscA4a4BY0QQtg8AiewQS8GU/Gq/FufMxaS0Yxsw/+lPH5BZIxpwo=</latexit><latexit sha1_base64="hP+6LrUf2d3tZaldqaQQvEKMXyw=">AAAB2XicbZDNSgMxFIXv1L86Vq1rN8EiuCozbnQpuHFZwbZCO5RM5k4bmskMyR2hDH0BF25EfC93vo3pz0JbDwQ+zknIvSculLQUBN9ebWd3b/+gfugfNfzjk9Nmo2fz0gjsilzl5jnmFpXU2CVJCp8LgzyLFfbj6f0i77+gsTLXTzQrMMr4WMtUCk7O6oyaraAdLMW2IVxDC9YaNb+GSS7KDDUJxa0dhEFBUcUNSaFw7g9LiwUXUz7GgUPNM7RRtRxzzi6dk7A0N+5oYkv394uKZ9bOstjdzDhN7Ga2MP/LBiWlt1EldVESarH6KC0Vo5wtdmaJNChIzRxwYaSblYkJN1yQa8Z3HYSbG29D77odBu3wMYA6nMMFXEEIN3AHD9CBLghI4BXevYn35n2suqp569LO4I+8zx84xIo4</latexit><latexit sha1_base64="WK6N/zm0v0B3ucyyuEhKiMuX03U=">AAACInicbZDLSsNAGIX/1FutVatbN4NFEISSuNGNIOjCZQV7gTaEyWTSDp1MwsxEGkJeyI2vIoKLivgsTi+Ktv4w8HHO3M7xE86Utu2JVVpb39jcKm9Xdqq7e/u1g2pbxakktEViHsuujxXlTNCWZprTbiIpjnxOO/7oZup3HqlULBYPOkuoG+GBYCEjWBvJq932I6yHfpiPC9QnQazRt5AV6AqNPWchZ4bO5qwMjD3xY5hr6nbDng1aBWcBdVhM06u99oOYpBEVmnCsVM+xE+3mWGpGOC0q/VTRBJMRHtCeQYEjqtx8lrZAJ0YJUBhLs4RGM/X3iRxHSmWRb3ZOo6hlbyr+5/VSHV66ORNJqqkg84fClCMdo2l1KGCSEs0zA5hIZv6KyBBLTLQpuGJKcJYjr0L7vOHYDefehjIcwTGcggMXcA130IQWEHiCF5jAu/VsvVkf87pK1qK3Q/gz1ucXa0ilcg==</latexit><latexit sha1_base64="WK6N/zm0v0B3ucyyuEhKiMuX03U=">AAACInicbZDLSsNAGIX/1FutVatbN4NFEISSuNGNIOjCZQV7gTaEyWTSDp1MwsxEGkJeyI2vIoKLivgsTi+Ktv4w8HHO3M7xE86Utu2JVVpb39jcKm9Xdqq7e/u1g2pbxakktEViHsuujxXlTNCWZprTbiIpjnxOO/7oZup3HqlULBYPOkuoG+GBYCEjWBvJq932I6yHfpiPC9QnQazRt5AV6AqNPWchZ4bO5qwMjD3xY5hr6nbDng1aBWcBdVhM06u99oOYpBEVmnCsVM+xE+3mWGpGOC0q/VTRBJMRHtCeQYEjqtx8lrZAJ0YJUBhLs4RGM/X3iRxHSmWRb3ZOo6hlbyr+5/VSHV66ORNJqqkg84fClCMdo2l1KGCSEs0zA5hIZv6KyBBLTLQpuGJKcJYjr0L7vOHYDefehjIcwTGcggMXcA130IQWEHiCF5jAu/VsvVkf87pK1qK3Q/gz1ucXa0ilcg==</latexit><latexit sha1_base64="UewclmH/JoUSEJcJsv9wiT4A0pI=">AAACLXicbZBNS8MwGMdTX+d8m3r0EhyCIIzWi16EoR48TnAvsJaSpukWlqYlSWWl9At58auI4GEiXv0aplsV3Xwg8OP/f54kz9+LGZXKNCfG0vLK6tp6ZaO6ubW9s1vb2+/IKBGYtHHEItHzkCSMctJWVDHSiwVBocdI1xtdF373gQhJI36v0pg4IRpwGlCMlJbc2o0dIjX0gmycQxv7kYLfQprDSzh2rVJONZ3OWGoYu/zH0NfUzYY5LbgIVgl1UFbLrb3YfoSTkHCFGZKyb5mxcjIkFMWM5FU7kSRGeIQGpK+Ro5BIJ5tum8NjrfgwiIQ+XMGp+nsiQ6GUaejpzmIVOe8V4n9eP1HBhZNRHieKcDx7KEgYVBEsooM+FQQrlmpAWFD9V4iHSCCsdMBVHYI1v/IidM4altmw7sx686qMowIOwRE4ARY4B01wC1qgDTB4BM9gAt6MJ+PVeDc+Zq1LRjlzAP6U8fkFkPGnBg==</latexit><latexit sha1_base64="tjPm2BYBzcpGEXsTQhjwuSSRkxI=">AAACLXicbZBdS8MwFIbT+TXn19RLb4JDEITRiqA3wlAvvJzgPmAtJU3TLSxNS5LKStkf8sa/IoIXE/HWv2G6VdHNA4GH9z0nyXm9mFGpTHNilJaWV1bXyuuVjc2t7Z3q7l5bRonApIUjFomuhyRhlJOWooqRbiwICj1GOt7wOvc7D0RIGvF7lcbECVGf04BipLTkVm/sEKmBF2SjMbSxHyn4LaRjeAlHrlXIqaaTGUsNI5f/GPqamlk3pwUXwSqgBopqutUX249wEhKuMENS9iwzVk6GhKKYkXHFTiSJER6iPulp5Cgk0smm247hkVZ8GERCH67gVP09kaFQyjT0dGe+ipz3cvE/r5eo4MLJKI8TRTiePRQkDKoI5tFBnwqCFUs1ICyo/ivEAyQQVjrgig7Bml95EdqndcusW3dntcZVEUcZHIBDcAwscA4a4BY0QQtg8AiewQS8GU/Gq/FufMxaS0Yxsw/+lPH5BZIxpwo=</latexit><latexit sha1_base64="tjPm2BYBzcpGEXsTQhjwuSSRkxI=">AAACLXicbZBdS8MwFIbT+TXn19RLb4JDEITRiqA3wlAvvJzgPmAtJU3TLSxNS5LKStkf8sa/IoIXE/HWv2G6VdHNA4GH9z0nyXm9mFGpTHNilJaWV1bXyuuVjc2t7Z3q7l5bRonApIUjFomuhyRhlJOWooqRbiwICj1GOt7wOvc7D0RIGvF7lcbECVGf04BipLTkVm/sEKmBF2SjMbSxHyn4LaRjeAlHrlXIqaaTGUsNI5f/GPqamlk3pwUXwSqgBopqutUX249wEhKuMENS9iwzVk6GhKKYkXHFTiSJER6iPulp5Cgk0smm247hkVZ8GERCH67gVP09kaFQyjT0dGe+ipz3cvE/r5eo4MLJKI8TRTiePRQkDKoI5tFBnwqCFUs1ICyo/ivEAyQQVjrgig7Bml95EdqndcusW3dntcZVEUcZHIBDcAwscA4a4BY0QQtg8AiewQS8GU/Gq/FufMxaS0Yxsw/+lPH5BZIxpwo=</latexit><latexit sha1_base64="tjPm2BYBzcpGEXsTQhjwuSSRkxI=">AAACLXicbZBdS8MwFIbT+TXn19RLb4JDEITRiqA3wlAvvJzgPmAtJU3TLSxNS5LKStkf8sa/IoIXE/HWv2G6VdHNA4GH9z0nyXm9mFGpTHNilJaWV1bXyuuVjc2t7Z3q7l5bRonApIUjFomuhyRhlJOWooqRbiwICj1GOt7wOvc7D0RIGvF7lcbECVGf04BipLTkVm/sEKmBF2SjMbSxHyn4LaRjeAlHrlXIqaaTGUsNI5f/GPqamlk3pwUXwSqgBopqutUX249wEhKuMENS9iwzVk6GhKKYkXHFTiSJER6iPulp5Cgk0smm247hkVZ8GERCH67gVP09kaFQyjT0dGe+ipz3cvE/r5eo4MLJKI8TRTiePRQkDKoI5tFBnwqCFUs1ICyo/ivEAyQQVjrgig7Bml95EdqndcusW3dntcZVEUcZHIBDcAwscA4a4BY0QQtg8AiewQS8GU/Gq/FufMxaS0Yxsw/+lPH5BZIxpwo=</latexit><latexit sha1_base64="tjPm2BYBzcpGEXsTQhjwuSSRkxI=">AAACLXicbZBdS8MwFIbT+TXn19RLb4JDEITRiqA3wlAvvJzgPmAtJU3TLSxNS5LKStkf8sa/IoIXE/HWv2G6VdHNA4GH9z0nyXm9mFGpTHNilJaWV1bXyuuVjc2t7Z3q7l5bRonApIUjFomuhyRhlJOWooqRbiwICj1GOt7wOvc7D0RIGvF7lcbECVGf04BipLTkVm/sEKmBF2SjMbSxHyn4LaRjeAlHrlXIqaaTGUsNI5f/GPqamlk3pwUXwSqgBopqutUX249wEhKuMENS9iwzVk6GhKKYkXHFTiSJER6iPulp5Cgk0smm247hkVZ8GERCH67gVP09kaFQyjT0dGe+ipz3cvE/r5eo4MLJKI8TRTiePRQkDKoI5tFBnwqCFUs1ICyo/ivEAyQQVjrgig7Bml95EdqndcusW3dntcZVEUcZHIBDcAwscA4a4BY0QQtg8AiewQS8GU/Gq/FufMxaS0Yxsw/+lPH5BZIxpwo=</latexit><latexit sha1_base64="tjPm2BYBzcpGEXsTQhjwuSSRkxI=">AAACLXicbZBdS8MwFIbT+TXn19RLb4JDEITRiqA3wlAvvJzgPmAtJU3TLSxNS5LKStkf8sa/IoIXE/HWv2G6VdHNA4GH9z0nyXm9mFGpTHNilJaWV1bXyuuVjc2t7Z3q7l5bRonApIUjFomuhyRhlJOWooqRbiwICj1GOt7wOvc7D0RIGvF7lcbECVGf04BipLTkVm/sEKmBF2SjMbSxHyn4LaRjeAlHrlXIqaaTGUsNI5f/GPqamlk3pwUXwSqgBopqutUX249wEhKuMENS9iwzVk6GhKKYkXHFTiSJER6iPulp5Cgk0smm247hkVZ8GERCH67gVP09kaFQyjT0dGe+ipz3cvE/r5eo4MLJKI8TRTiePRQkDKoI5tFBnwqCFUs1ICyo/ivEAyQQVjrgig7Bml95EdqndcusW3dntcZVEUcZHIBDcAwscA4a4BY0QQtg8AiewQS8GU/Gq/FufMxaS0Yxsw/+lPH5BZIxpwo=</latexit><latexit sha1_base64="tjPm2BYBzcpGEXsTQhjwuSSRkxI=">AAACLXicbZBdS8MwFIbT+TXn19RLb4JDEITRiqA3wlAvvJzgPmAtJU3TLSxNS5LKStkf8sa/IoIXE/HWv2G6VdHNA4GH9z0nyXm9mFGpTHNilJaWV1bXyuuVjc2t7Z3q7l5bRonApIUjFomuhyRhlJOWooqRbiwICj1GOt7wOvc7D0RIGvF7lcbECVGf04BipLTkVm/sEKmBF2SjMbSxHyn4LaRjeAlHrlXIqaaTGUsNI5f/GPqamlk3pwUXwSqgBopqutUX249wEhKuMENS9iwzVk6GhKKYkXHFTiSJER6iPulp5Cgk0smm247hkVZ8GERCH67gVP09kaFQyjT0dGe+ipz3cvE/r5eo4MLJKI8TRTiePRQkDKoI5tFBnwqCFUs1ICyo/ivEAyQQVjrgig7Bml95EdqndcusW3dntcZVEUcZHIBDcAwscA4a4BY0QQtg8AiewQS8GU/Gq/FufMxaS0Yxsw/+lPH5BZIxpwo=</latexit>

@willb

Other technologies

def dot(xs, ys): return sum([x * y for x, y in zip(xs, ys)])

@willb

Other technologies

dot([0.1, 0.2, 0.3], [1.0, 2.0, 3.0])

@willb

0.1d 0.2d 1.0d 2.0d

0.3d (unused) 3.0d (unused)

@willb

0.1d 0.2d 1.0d 2.0d

0.3d (unused) 3.0d (unused)

vdpps0.3f unused0.2f0.1f 3.0f unused2.0f1.0f

@willb

Other technologies

@willb

Other technologies

@willb

Other technologies

@willb

Other technologies

@willb

Other technologies

@willb

Ecosystem demos at Summit

What did we talk about today?

@willb

https://www.redhat.com

THANKS

openshift for ai and ml -...

Documents

openshift enterprise

openshift container platform 3.7 architecture · openshift...

openshift docker, kubernetes and … automation using...

openshift 4 update

openshift learning path - redhat.com · learning path...

openshift workshop

openshift container platform 3.9 architecture · openshift...

openshift for continuous integration alex lossent –...

using openshift paas

openshift 4 update - red...

openshift online 3 developer guide · openshift online 3...

openshift experience @ puzzle · 1 2013-06-11 / open cloud...

kubernetes - openshift

all things openshift

release update red hat openshift 4 -...

openshift express intro

openshift container platform 3 - red...

openshift dedicated 4 networking€¦ · openshift...

openshift container platform 4.1 applications...2020/03/19...

guide for running ai workloads on red hat openshift and...