Discovering AI Innovations on Google Distributed Cloud - 蔡秀吉

Slide 1

Slide 1 text

Google Distributed Cloud (GDC) Discover AI innovations and modernization @thc1006

Slide 2

Slide 2 text

Google Cloud Proprietary & Confidential Google Distributed Cloud (GDC) GDC AI/ML Gemini Cloud Assist 01 02 03 04 Agenda 2

Slide 3

Slide 3 text

Why not use Public Cloud for everything? AI Section 1

Slide 4

Slide 4 text

“ ” Mark Lohmeyer

Slide 5

Slide 5 text

AI model

Slide 6

Slide 6 text

5G/Wi-Fi CPE • • • Public Cloud video surveillance

Slide 7

Slide 7 text

Public Cloud

Slide 8

Slide 8 text

No content

Slide 9

Slide 9 text

No content

Slide 10

Slide 10 text

No content

Slide 11

Slide 11 text

Google Cloud Proprietary & Confidential /

Slide 12

Slide 12 text

⚫ ⚫ ⚫ ⚫

Slide 13

Slide 13 text

Section 1

Slide 14

Slide 14 text

From edge, to sovereign, to cross-cloud What is Google Distributed Cloud? Section 2

Slide 15

Slide 15 text

Google Cloud Proprietary & Confidential Edge (On premises ( )Air Gap Google Cloud

Slide 16

Slide 16 text

Google Distributed Cloud

Slide 17

Slide 17 text

17 Configurations Air-Gapped Connected

Slide 18

Slide 18 text

GenAI Apps Use cases on Google Distributed Cloud Google Distributed Cloud Technology Stack Section 2

Slide 19

Slide 19 text

No content

Slide 20

Slide 20 text

No content

Slide 21

Slide 21 text

No content

Slide 22

Slide 22 text

GDC enables the latest technology • Performance Tuning Operator • SR-IOV support for NICs • Multus - multiple NICs per container • Direct resource access (CPU and memory) • On-prem Nodes with 128 CPUs and 4TB of RAM • Support up to 30,000 pods per cluster • NVIDIA GPU support • Observability

Slide 23

Slide 23 text

NVIDIA L4 GPU NVIDIA H100 GPU on prem

Slide 24

Slide 24 text

Google Cloud Proprietary & Confidential

Slide 25

Slide 25 text

No content

Slide 26

Slide 26 text

Slide 27

Slide 27 text

GDC Service Platform & IaaS Service Platform(Full Cloud-Native Stack) (observability) (hardening) Dashboard Network fabric topology Spanning from IAM, monitoring, logging, management Sys IaaS Stack: GKE Enterprise load balancer Life Cycle Management (LCM) https://cloud.google.com/blog/products/management-tools/introducing-customizable-observability-dashboards/

Slide 28

Slide 28 text

GDC PaaS Stack ⚫ AI AlloyDB GDC Vector search ⚫ GDC Dataproc GDC

Slide 29

Slide 29 text

Google Cloud Proprietary & Confidential Vertex AI Search & Prediction AlloyDB Omni Dataproc Spark

Slide 30

Slide 30 text

Google Cloud Proprietary & Confidential AI Stack GDC Workbench Data Pipelines AI AI

Slide 31

Slide 31 text

Google Cloud Proprietary & Confidential Migration GDC GDC MongoDB, Elasticsearch, etc,.

Slide 32

Slide 32 text

AI/ML on Google Distributed Cloud (GDC) Section 3

Slide 33

Slide 33 text

Google Cloud Proprietary & Confidential DEMO

Slide 34

Slide 34 text

No content

Slide 35

Slide 35 text

No content

Slide 36

Slide 36 text

GDC Apigee API

Slide 37

Slide 37 text

From edge, to sovereign, to cross-cloud Cross-CIoud Network Section 2

Slide 38

Slide 38 text

No content

Slide 39

Slide 39 text

GenAI Apps Use cases on Google Distributed Cloud Gemini Cloud Assist Section 2

Slide 40

Slide 40 text

DEMO

Slide 41

Slide 41 text

41 Configuration-as-Data is Declarative (Not Imperative) I’m thirsty… I want soda in my blue cup. Sure! Thanks for trusting me to figure out how to get you a soda. Express intent… ...without the toil of “how” Declarative: User Friendly Imperative : I’m thirsty… Go to the kitchen Open the fridge Reach in the back to get the soda Close the fridge Grab a blue cup from the left cupboard Pour out the soda in the cup Bring the cup back to me Ummm… okay thanks for telling me exactly what to do...sheesh!