Fundamentals of Google Cloud: A Guided Tour

Slide 1

Slide 1 text

Conﬁdential & Proprietary Fundamentals of Google Cloud: A Guided Tour Florian Rathgeber Site Reliability Engineer at Google Cloud @frathgeber follow me on twitter! Slide credits:

Slide 2

Slide 2 text

Florian Site Reliability Engineer Google Cloud ● ● ● ●

Slide 3

Slide 3 text

No content

Slide 4

Slide 4 text

Proprietary + Conﬁdential

Slide 5

Slide 5 text

Proprietary + Conﬁdential

Slide 6

Slide 6 text

Compute Running your code

Slide 7

Slide 7 text

Virtual Machines Containers Serverless

Slide 8

Slide 8 text

No content

Slide 9

Slide 9 text

Highly customisable Highly managed Compute Engine Cloud Marketplace App Engine Cloud Build Container Registry Kubernetes Engine Cloud Functions Cloud Run

Slide 10

Slide 10 text

When you use Google Cloud, you’re using the network that powers... Seven cloud products with ONE BILLION users each

Slide 11

Slide 11 text

Edge points of presence >100 Network sea cable investments Google global cache edge nodes (>800) Network

Slide 12

Slide 12 text

Storage Storing your data

Slide 13

Slide 13 text

Where do I store my data? In Memory Relational NoSQL Object Block Warehouse Cloud Memorystore Good for: Web/mobile apps, gaming Such as: Game state, user sessions Cloud SQL Good for: Web frameworks Such as: CMS, eCommerce Cloud Spanner Good for: RDBMS+scale, HA, HTAP Such as: User metadata, Ad/Fin/MarTech Cloud Datastore Good for: Hierarchical, mobile, web Such as: User proﬁles, key/val data Cloud Bigtable Good for: Heavy read + write, events Such as: AdTech, Financial, IoT Cloud Storage Good for: Binary or obj data (BLOB) Such as: Images, media, archive, backup Persistent Disk (GCE) Good for: Local VM ﬁle storage Such as: App data & binaries Big Query Good for: Data Warehouse Such as: Analytics, dashboards

Slide 14

Slide 14 text

Structured Data (Relational) Cloud SQL Spanner Proprietary + Conﬁdential

Slide 15

Slide 15 text

Unstructured Data (NoSQL) Datastore Firestore BigTable Proprietary + Conﬁdential Photo by Sebas Ribas on Unsplash

Slide 16

Slide 16 text

Object Store (Blobs) Cloud Storage Proprietary + Conﬁdential

Slide 17

Slide 17 text

Proprietary + Conﬁdential

Slide 18

Slide 18 text

Big Data Extracting Insights

Slide 19

Slide 19 text

BigTable Dremel Colossus Flume Megastore Spanner PubSub Millwheel Cloud Dataﬂow Cloud Dataproc MapReduce

Slide 20

Slide 20 text

Big Data Lifecycle Google Analytics Premium Cloud Pub/Sub Capture Google Stackdriver Firebase Storage Transfer Service Use Data Scientists Business Analysts Cloud Datalab ... BigQuery Storage (tables) Cloud Bigtable (NoSQL) Cloud Storage (files) Store Cloud Dataflow BigQuery Analytics Analyze Cloud Dataproc ML Engine SQL Process Stream Real-time analytics Real-time dashboard Real-time alerts Batch Cloud Dataflow Cloud Dataprep

Slide 21

Slide 21 text

Machine Learning Learning from your data

Slide 22

Slide 22 text

Accelerating use of deep learning at Google Directories containing Deep Learning Models 2012 2013 2014 2015 3000 2000 1000 0 Used across products: 4000 2016 Unique project directories

Slide 23

Slide 23 text

Use our pre-trained models Create & serve your custom models

Slide 24

Slide 24 text

No content

Slide 25

Slide 25 text

● ● ● ● Learn more: What is the Tensor Flow machine intelligence platform?

Slide 26

Slide 26 text

Framework GitHub Star Count TensorFlow 45581 scikit-learn 16423 Caffe 15882 CNTK 9540 MXNet 8226 Torch 6344 Theano 5647 2010 2013 2015 2014 2016 2017 12500 25000 37500 50000 0

Slide 27

Slide 27 text

No content

Slide 28

Slide 28 text

Train your model with ML Engine trainingInput: scaleTier: CUSTOM masterType: complex_model_m workerType: complex_model_m parameterServerType: large_model workerCount: 9 parameterServerCount: 3 gcloud ml-engine jobs submit training $JOB_NAME \ --package-path $TRAINER_PACKAGE_PATH \ --module-name $MAIN_TRAINER_MODULE \ --job-dir $JOB_DIR \ --region $REGION \ --config config.yaml \

Slide 29

Slide 29 text

Monitor ML Engine job on Cloud Console

Slide 30

Slide 30 text

Slide 31

Slide 31 text

No content

Slide 32

Slide 32 text

© 2018 Google LLC. All rights reserved. An Algorithm that can create a Machine Learning Model Cloud AutoML allowing your developers to create high quality custom Models Introducing Cloud AutoML Human Data Labeling: For customers with no labeled training images, our in-house human labelers are available to review your custom instructions and label your images accordingly for model training. Production Ready: Use it immediately, Cloud AutoML is already deployed in GCP, Scales well with GCP large scale computation resource

Slide 33

Slide 33 text

Slide 34

Slide 34 text

Resources Main page cloud.google.com Console console.cloud.google.com Codelabs g.co/codelabs/cloud Training cloud.google.com/training

Slide 35

Slide 35 text

cloud.google.com/free

Slide 36

Slide 36 text

Conﬁdential & Proprietary Thank you! Florian Rathgeber @frathgeber follow me on twitter!