spotinst 'aws cost optimization' webinar - jan 20th, 2016

Post on 15-Apr-2017

390 Views

Category:

Technology

2 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Saving Thousands By Running Mission Critical Applications on Spot

January 20th, 2016

© 2016 Spotinst, LTD. #spotinstmeetup

Welcome!

To our 1st online

• Questions are always welcome• There are no bad questions• We’ll demo as much as we can• Things will break, be patient

Amiram ShacharFounder & CEO / Spotinst

Gal AvivR&D Group Manager /

Inneractive

Shannon WilliamsCo-Founder & VP Marketing /

Rancher Labs

Shimon ToltsHead of Infrastructure /

ironSource

@AmiramShachar @galaviv @shimtolts @smw355

Agenda• The Basics: Spotinst on AWS• Can I run my application on Spot?• Demo: Spotinst floating IPs, Beanstalk, CloudFormation• Use-case: Inneractive & Spotinst, Gal Aviv• Use-case: ironSource & Spotinst, Shimon Tolts• Using Rancher for Docker production workloads, Shannon• Demo: NEW! Spotinst & EMR autoscaling

The ultimate platform for running and managing clusters in the cloud.

Spot = Idle On Demand

Spot is a market where the price of compute changes based on supply and demand

The Basics

Spot Strategy #1Significantly reduce your EC2 costs

cloud expenses graph of one of Spotinst’s customers. Before and after using Spotinst

Budget EC2 Capacity Purchased

Spot Strategy #2Grow your compute capacity and throughput 10x for the same budget.

Spot Barriers

SLA Capacity is not guaranteed.

Safe application drainingWhat happens if AWS pulls the plug?

The Spotinst Way

Jul August September Octber November DecemberRunning Hours of Spotinst Managed Cloud Servers

Spotinst traffic usageIncreased 100% MoM

1,000,000

What can run on Spot?

Everything. except single point of failures.

Spot Ready

Hadoop / Spark

ELB / HA-Proxy Docker / ECS Autoscaling

MicroservicesWeb Application

Demo: Safe Draining

Demo: Floating IPs, Beanstalk, CloudFormation

© 2015 Rancher Labs, Inc.© 2015 Rancher Labs, Inc .

Rancher Labs OverviewShannon WilliamsJanuary 2016

© 2015 Rancher Labs, Inc.

Public container services are becoming popular

Google Container Engine

© 2015 Rancher Labs, Inc.

New private cloud: private container service

Private Container Service

Linux Linux Linux Linux Linux Linux

© 2015 Rancher Labs, Inc.

How to build a private container service

Private Container Service

Linux Linux Linux Linux Linux Linux

Docker MachineSwarm, Kubernetes, MesosCompose, Kubernetes, Marathon, TOSCARancherLDAP, AD, GitHubStats, Health checks, Sysdig, Datadog, …Notary, Kubernetes, VaultVXLAN, IPSECiptablesHAProxy, iptables, ngnixCeph, EBS, SAN, ...Gluster, EFS, NAS, ...Swift, S3, ...etcd, consul, mongo, DynamoDB, ...

Resource MgmtSchedulingOrchestrationApp CatalogAccess ControlMonitoringSecurityOverlay networkFirewallLoad BalancerBlock StorageFile StorageObject StorageDistributed DB

© 2015 Rancher Labs, Inc.

Open source container management platform for building a Private Container Service

A containerized Linux distribution for running Docker

© 2015 Rancher Labs, Inc.

Rancher OverviewRancher Catalog

AD/LDAP

Alerts/Events

Environment 1 Environment N…Network Services Storage Services

Container Orchestration and SchedulingUser Mgmt Ops Mgmt

Environment 2

© 2015 Rancher Labs, Inc.

• Leverage spot instances to reduce infrastructure costs • Identify upcoming changes and orchestrate reaction• Working on live migration for 2016

G a l A v i vR & D G r o u p M a n a g e r

Spotinst Case Study

What we do…

• Real time trading and optimizing campaigns over ad spaces

• Real time analytics• Machine learning and recommendations• Trading data and insights• Our tech stack: Java, Scala, AKKA, Spray, Kafka

Spark, Spark Streaming and Cassandra

Scale and numbers…

• Billions of daily incoming requests

• Tens of Billions of daily transactions

• 150ms average response time

• Hundreds of servers in dozens of elastic clusters

• Extreme scaling (up/down)

Scaling challenges

• Different cluster behavior (latency, CPU, load, QPS…)

• Tight back-pressure management (Scala/AKKA/Spray stack)

• Critical serving core (99.99% uptime)

• Service cost efficiency is crucial

Spotinst integration

• Less than 1 day for initial setup

• Cluster by cluster partial migration

• Extreme costs reduce

• Cost vs. risk… Series10%

25%

50%

75%

100%

Reserved Spotinst OnDemand

Nice to meet you!Shimon ToltsHead of infrastructure @ ironSource

ironBeastData Flow Management Solution

SpotInst case study Shimon Toltshttp://www.ironsrc.com/ironbeast/

PEOPLE REACHED EACH MONTH APPS INSTALLED EVERY MINUTEWITH THE IRONSOURCE PLATFORM

REGISTERED & ANALYZED DATA EVENTSEVERY MONTH

600M 4,200

100B+

70K+PARTNERED APPS

SEP. 2010ESTABLISHED

3.5B+INSTALLATIONS TO DATE

700TOTAL EMPLOYEES

400R&D EMPLOYEES

IRONSOURCE IN NUMBERS

TEL AVIV ISRAEL SAN FRANCISCO UNITED STATES NEW YORK UNITED STATES LONDON UNITED KINGDOM BEIJING CHINA

ironBeast

Simplifying Big Data Flow Management

Learn more >

● Our deployment - micro-services with Docker

● Implement it yourself?● 4 Availability Zones ● 5 Auto Scaling groups? (1 on demand)● c4.large, c3.large, r3.large, m4.large● 4x4=16 ASG + 1 On demand

SPOT INSTANCES

Spotinst

● One place to configure it all● Multi instance type support● Deep visibility into cost● High savings● Easy Risk/Reward mitigation (% of spot)

● Spot clean up mechanism

Thank you

See you in our next webinar!

@Spotinst /spotinstance/spotinst

NEW! EMR Task Nodes AutoScale

can run applications on Amazon’s and Google’s cloud for 1/10th of the regular price

It is a fact

top related