Modernization in a

Hybrid/Multi-Cloud
World Nick Ihli
Google Cloud
Customer Engineer - Utah

Nick Ihli
#FlowFactorSLC
@IhliNick
 Scalability and
reliability is a part of
our DNA
8 products
with > 1BN users

Nick Ihli
#FlowFactorSLC
@IhliNick
 Better global network infrastructure
A privately-owned network, isolated from the public internet

3
FASTER (US, JP, TW) 2016
Havfrue (US,IE, DK) 2019
3 3
3 3
3 3 3
4

3 3 3
3 3
3
3
3 3 SJC (JP, HK, SG) 2013

HK-G (HK, GU) 2019

Unity (US, JP) 2010
Curie (CL, US) 2019
2 PLCN (HK, LA) 2019

Current region and number of zones

3 Monet (US, BR) 2017

3 Future region and number of zones

Junior (Rio, Santos) 2017
Network
Tannat (BR, UY, AR) 2017

Edge points of presence >100

Indigo (SG, ID, AU) 2019 Edge node locations >1000

Nick Ihli
#FlowFactorSLC
@IhliNick
At Google, everything
runs in a container

On average, we launch

4 billion
new containers per week

Nick Ihli
#FlowFactorSLC
@IhliNick
 15+ Years of Innovation

Open
Source
Apache Beam

Google Map Flume

GFS BigTable Dremel PubSub Millwheel Tensorflow
Research Reduce Java

Google
Cloud
Products BigQuery Pub/Sub Dataflow Bigtable Dataproc ML

2002 2004 2005 2006 2008 2010 2012 2014 2015 2016

Nick Ihli
#FlowFactorSLC
@IhliNick
 Announced
GKE on Prem
Timeline of Kubernetes

GKE GA- Aug 15

Work begins to

Envoy 1.0
opensource Google’s
Public Beta of Solaris Borg as Kubernetes

Knative
Containers

2003 2006 2009 2012 2015 2018

Docker Launched 2013 GKE Alpha -

Kubernetes 1.0 - July 15

Process Containers
Nov 14

Gvisor
launched by Google now

Istio 0.1

Istio 1.0
Birth of Borg, 3-4 Google known as cgroups and
Engineers working to merged with Linux
automate cluster Kernel. LXC launched, complete
management. First Github commit
Linux container manager
for Kube - Jun 14

Nick Ihli
#FlowFactorSLC
@IhliNick
6 transformational technologies
Open software Microservices Containerization

● Move at the speed of your

● Avoid lock-in ● Increase portability of
fastest developer
● Avoid unnecessary applications
● Use the best tools
licensing costs ● Increase security
● Leverage small and agile
● Broad ecosystem of partners ● Increase speed
teams

CI/CD and DevOps Orchestration Services and API management

● Define goals, instead of ● Decrease downtime

● Create a software
actions ● Increase portability of
assembly line
● Less idle resources services
● Build fast feedback into your
● Manage apps instead of ● Manage access and testing
systems
servers of services

Nick Ihli
#FlowFactorSLC
@IhliNick
 Site Reliability Engineering

https://landing.google.com/sre/

Nick Ihli
#FlowFactorSLC
@IhliNick
Nick Ihli
#FlowFactorSLC
@IhliNick
Nick Ihli
#FlowFactorSLC
@IhliNick
Enterprise IT is complex and
expensive

Cloud is Hybrid is key But most apps Complexity

here to stay Multi-cloud too remain on-prem & Cost

73% 84% 10% 83%

of organizations Of companies have Workloads have increase in
use cloud1 multi-cloud plans2 moved to cloud3 administration costs4

1
IDG Cloud Insights survey, 2018 (N=550)
2
Cloudability, State of Cloud 2018 report based on actual spend data
4
IDC: study
Nick Ihli
#FlowFactorSLC
@IhliNick
Enterprise IT
is complex

Enterprise IT is Businesses
complex demand agility

Nick Ihli
#FlowFactorSLC
@IhliNickv
IT is forced into trade-offs

Security vs Agility Reliability vs Cost Portability vs Consistency

● Development wants to ● Lock-in gives you

● Redundancy is the
use the latest tools, consistency
way to minimize
languages and plugins
downtime
● Lock-in keeps you from
● Security wants verified using the best tools,
● Redundancy means
tools and validated negotiating better
underutilized
systems before putting prices, or being as
resources
it into production flexible as you want

Nick Ihli
#FlowFactorSLC
@IhliNick
Hybrid done right
A software-based stack
means no hardware purchase required.
Zero to deployed in less than a few hours
rather than months.

Built on open software

for uniformity; one platform that can run both
on-prem and in the cloud without lock-in

Infrastructure abstracted away

focus on building apps not managing
infrastructure.

Nick Ihli
#FlowFactorSLC
@IhliNick
Introducing Google Cloud’s
Anthos
Google Cloud All Major Clouds

Anthos lets you build and manage modern

hybrid and multi-cloud applications without
lock-in

Build once, to run anywhere, across your

existing on-premise infrastructure and all
major public cloud providers

On-Premise Data Center

Nick Ihli
#FlowFactorSLC
@IhliNick
Anthos: Bringing the cloud to you.
Solutions

Serverless CI/CD

Core Services
On-prem Marketplace

Cloud

Service Logging &

management monitoring

Config.
management

Nick Ihli
#FlowFactorSLC
@IhliNick
Transform your IT and build apps for today
and the future with Anthos

Modernize in-place Automate policy and Consistency

security at scale
Modernize your applications no Proactive service operations - Anthos gives you one platform that you
matter where they are. Consistent manage at a higher layer of the can run anywhere. It’s built on open
management of your applications stack, enabling greater application source technology created by Google;
across multiple clouds and awareness, consistency, and so it’s portable, consistent, and
on-premises. Faster time to market, control. Take a service-centric view extensible to help you future-proof your
lower administrative overhead, and of your infrastructure. investments.
increased innovation capabilities.

Nick Ihli
#FlowFactorSLC
@IhliNick
 New
Anthos: A hybrid and multi-cloud platform to build once run anywhere
What’s available

Google Cloud Platform On-Premises Other Public Cloud

AWS, Azure, and other clouds

Cloud Run Cloud Run

GCP Marketplace GCP Marketplace

Cloud Service Mesh

Anthos Config
Anthos Config Management
Management Available

GKE GKE On-Prem Other Kubernetes Cluster Announced

Core
Infrastructure
On-Prem Infrastructure GCP Infrastructure
 Get Started Reliable and Integrated with
Quickly Available GCP

Use the Cloud GKE Take advantage

Console to automatically of the deep
easily create a repairs, integration with
Google cluster and view upgrades, and Google Cloud

Kubernetes Engine your workloads scales your

cluster with a
Platform
features and
(GKE) high-availability services
control plane

20
Nick Ihli
#FlowFactorSLC
@IhliNick
 Turn-key, production-grade, Easy upgrade path to the
conformant Kubernetes with latest Kubernetes releases
best-practice configuration that have been validated and
tested by Google

GKE On-Prem
Access to container services Integration with Istio, Cloud
on GCP such as Cloud Build, Run, marketplace solutions
Container Registry, Audit
Logging, and more

Nick Ihli
#FlowFactorSLC
@IhliNick
 vCenter APIs
vSphere

Admin Cluster

GKE On-Prem
Virtual Machine

Admin Workstation
Cluster creation, kubelet
upgrade, teardown
gkectl Admin Control
Plane

kubectl
● Automated deploy on top of vSphere User Cluster User Cluster
shipped as a virtual appliance Virtual Machine Virtual Machine

● Admin cluster provides local CRUD kubelet kubelet

operations: User Control Plane User Control Plane

Faster, higher reliability, easier to

manage from cloud
Virtual Machine Virtual Machine Virtual Machine

Independent failure domain kubelet kubelet kubelet

● Approach extensible to other IaaS Workload Workload Workload

F5 BIG-IP LTM
Service Mesh (Istio)
Uniform Operational Policy driven
observability agility Security
What does it do?

Connect, secure, manage,

monitor services
Anthos Config Multi-cluster management Hybrid support

Management Central Git repository

manages access control
Manages both on-prem and in
the cloud; change central file
policies, resource quotas, and apply across the fleet
A single pane of glass for managing
and namespaces
configurations and policies both
on-prem and in the cloud
Declarative and continuous Simple migration

Declare new desired state, Uses YAML or JSON, so no

continuously checks for rewriting of existing
changes that go against state Kubernetes configs

Nick Ihli
#FlowFactorSLC
@IhliNick
Single pane of glass
across GCP & on-prem

Orchestrate and manage on-prem containers

just like GKE in the cloud

Consistent operating model with access to GCP

services across hybrid environments

Single-pane-of-glass for multiple Kubernetes

clusters, no matter where

Nick Ihli
#FlowFactorSLC
@IhliNick
Upgrade VMs/servers directly into containers in
GKE

Anthos Migrate BETA ● Migrate & modernize VMs from on-prem,

GCE, or other clouds directly into
containers in GKE.
● Introduces a “lift & modernize” play.
● Built-in testing
● Minimal labor, downtime, complexity

Nick Ihli
#FlowFactorSLC
@IhliNick
Why use
Kubernetes applications in
Google Cloud Marketplace?
Run anywhere
On premises, Anthos, or to
Kubernetes clusters hosted in other
environments

Be more productive
Prebuilt deployment templates and
default configurations, click to
deploy to GKE, and managed
upgrades (alpha)

Simplify billing
Usage metering follows the
application no matter where it
is deployed. Consolidated with
other GCP spend

Nick Ihli
#FlowFactorSLC
@IhliNick
 New
Anthos: A hybrid and multi-cloud platform to build once run anywhere
What’s available

Google Cloud Platform On-Premises Other Public Cloud

AWS, Azure, and other clouds

Cloud Run Cloud Run

GCP Marketplace GCP Marketplace

Cloud Service Mesh

Anthos Config
Anthos Config Management
Management Available

GKE GKE On-Prem Other Kubernetes Cluster Announced

Core
Infrastructure
On-Prem Infrastructure GCP Infrastructure
That’s a wrap.

29