Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

CMP325-R

Using Amazon EBS to build highly


resilient applications
Paul Fryer Peter M. O'Donnell
Principal Solutions Architect Security Specialist Solutions Architect
Amazon Web Services Amazon Web Services

© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Agenda
Amazon Elastic Block Store (Amazon EBS) overview

Common patterns for resiliency

Cloud-native patterns for stateful storage

New capabilities for Amazon EBS

CloudEndure
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Quick Amazon EBS overview & refresh
• Amazon Elastic Block Store
• EBS volumes & snapshots EC2
instance
• Virtual disks for virtual
computers
• Dependent on OS interaction
with block devices for storage
• Different types of volumes for EBS EBS EBS
different types of workloads volume
(boot)
volume
(data)
volume
(data)
Availability Zone

AWS Region
Matching Amazon EBS performance to Amazon EC2
performance
• EBS volume performance is
dependent on the EC2 instance
• EC2 families and sizes have
different aggregate EBS
capacities gp2 io1
General purpose SSD Provisioned IOPS SSD
• Mismatches between volume
performance and instance
capability can lead to
unexpected behaviors
• Oversizing EBS is neither
performance efficient nor cost
optimized st1 sc1
Throughput-optimized HDD Cold HDD
Common patterns for resiliency
• File-level recovery demands file-
level backups
• Snapshots in-region for volume
failure
99.999% service availability
• Snapshot copy x-region for
regional durability
• Cautions about mirroring and
striping
• Multi-volume consistent
snapshots 0.1% to 0.2% annual
failure rate (AFR)
Best practice: RAID

When to RAID?
• Storage requirement >16 TiB

• Throughput requirement >1,000 MiB/s


EBS EBS EBS

• IOPS requirement >64,000 @ 16K


Best practice: RAID

Avoid RAID for redundancy


• Amazon EBS data is already replicated

• RAID1 halves available Amazon EBS bandwidth

EBS EBS EBS • RAID5/6 loses 20% to 30% of usable I/O to parity
What can you do with a snapshot?

EBS EC2 EBS EC2


volume instance volume instance

Availability Zone Availability Zone

EBS
snapshot

AWS Region
What can you do with a snapshot?

EBS EC2 EBS EC2


volume instance volume instance

Availability Zone Availability Zone

EBS EBS
snapshot snapshot
AWS Region AWS Region
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Common cloud-based disaster recovery drivers

Improve recovery Consolidate DR Start to build TCO


objectives to meet strategy for multiple cloud skill set reduction
business requirements applications

Threat of data loss due to Protect large-scale, DR site hardware DRaaS provider
ransomware and server compute-intensive refresh contract changes
corruption workloads
Cloud-based disaster recovery – Business outcomes

Robust operations
Achieve steadfast reliability and availability based on top-tier
recovery objectives

Operational efficiency
Obtain substantial cost savings by reducing the need for duplicate
infrastructure and licensing

Resilience peace of mind


Minimize downtime and data loss by conducting easy-to-launch,
nondisruptive disaster recovery tests
Common disaster recovery challenges

• Diverse infrastructure • Inadequate testing and drills


and OS types
• High cost of duplicate
• Machine compatibility issues infrastructure & licensing
• Amount of time required to • Different DR tools for different
recover workloads applications
• Busy, continually • Inability to achieve recovery
changing workloads objectives (RPOs/RTOs)
Better, faster, more affordable disaster recovery
Flexible Reliable Highly automated

• Improve recovery objectives & reduce TCO


Replicate from
Robust, predictable,
nondisruptive
Minimal skill set • Simple setup lets you start in minutes
any source required to operate
continuous replication
• Same highly automated process for all
workloads
Wide range of OS, RPO: Sub-second Easy, nondisruptive
• Common workloads include databases such as Microsoft
application, and RTO: Minutes DR tests SQL Server, Oracle, and MySQL and enterprise
database support applications such as SAP
• Minimizes complexity and reduces risk

Automated
• Easy failover and failback
Protection against
Failback to lightweight staging
ransomware,
cloud/on prem area reduces TCO
corruptions, and human
errors
How CloudEndure Disaster Recovery works
• CloudEndure continuously replicates any application or database from any source into AWS
• Business outcome: Use AWS to significantly reduce DR TCO while improving recovery objectives

Oracle
Database Oracle
Database

Microsoft Microsoft
SQL Server SQL Server
Wide platform support*

Any Any x86 operating Source


application database systems infrastructure

* See documentation or contact Support for a complete list


CloudEndure Disaster Recovery success stories
• After deploying CloudEndure Disaster Recovery, Malibu Boats experienced an outage
• Failover to AWS in recovery mode resulted in 2x faster performance with same machine specs
• Led to a business decision to significantly expedite migration to AWS

• Used CloudEndure to replicate hundreds of servers into low-cost DR site in AWS


• Automated orchestration of business-critical ERP applications built on Microsoft .NET, SQL
Server, IIS, and clustering technologies
• Reduced DR total cost of infrastructure by at least 50% while improving recovery objectives

Worldwide Flight • Needed to meet aggressive recovery targets for compliance with government security guidelines
• Deployed DR solution for all busy servers, including SQL and IBM MQ servers
Services (WFS) • Success with DR project prompted decision to migrate entire production environment to AWS

• Decided to shift DR strategy to AWS instead of refreshing old DR infrastructure


Transdev • Estimated that building DR in the cloud vs. on premises reduced costs by 73%
• Achieved aggressive RPOs and RTOs for critical MS workloads, including Active Directory, SQL
Servers, MSCS clustered SAP workloads, and niche applications
Disaster recovery options: Before or after migration

Pre-migration benefits Post-migration benefits


• Reduce IT resilience costs and • Increase resilience of
improve recovery with AWS migrated applications
as your DR target
• Leverage cross-region DR to
• Easily test production achieve geographic redundancy
workloads in the cloud and
speed up your familiarity with • Protect environment from data
AWS corruption, ransomware, or other
malicious attacks
• Set the stage to deploy a
seamless “one-click migration”
How to obtain CloudEndure licenses

• Purchase CloudEndure Disaster


Recovery on AWS Marketplace
with self-service setup
• Contact your consulting partner,
account team, SI, or managed
service provider
Thank you!

© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.

You might also like