OpenShift for AI and ML -...

OpenShift for AI and MLWilliam Benton • @willb • [email protected]

@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

@willb

f( ) = 0.67 0.57 0.84 0.08 0.42 0.01

@willb

@willb

data collection and cleaning

codifying problem and metrics

@willb

feature engineering

model training and tuning


@willb

feature engineering


model validation

@willb

model validation

model deployment

monitoring and validation

@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

%

@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

/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

55

@willb



@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

OK!

@willb


OK!base image


application code

@willb


base image


application code

What OpenShift offers working data scientists

@willb

workflows to accelerate discovery


@willb

@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

@willb

OK!

OK!

@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

@willb

mA

@willb

mA

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

transform

archive

@willbdata scientists

application developersdata engineers

federate

trainmodels

events

databases


management

web and mobile

reporting


transform

transform

archive


application developersdata engineers

federate

trainmodels

events

databases


management

web and mobile

reporting


transform

transform

archive

machine learning engineers

@willb

data engineers

federate

events

databases


transform

transform

transform

archive


federate

trainmodels


transform

transform

@willb

application developers

federate

trainmodels management

web and mobile

reporting


transform

transform

archive

@willb



@willb

Multitenant compute clustersCluster scheduler

Shared FS / object store

Compute node

Compute node

Compute node

Compute node

Compute node

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 3

app 6 Databases

@willb

One cluster per applicationOpenShift

Shared FS / object storeapp 1 app 2

app 5app 4

app 3

app 6

app 2

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

vdppd

vdppd

0.1d 0.2d 1.0d 2.0d

0.3d (unused) 3.0d (unused)


@willb

vdppd

vdppd

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

Ecosystem demos at Summit

What did we talk about today?

@willb

@willb

https://www.redhat.com

THANKS

OpenShift for AI and ML -...

Documents

Transcript of OpenShift for AI and ML -...