These days, EVERY workload is considered critical by someone in the organization. As a result, SLAs are shrinking. IT is challenged to meet these SLAs, but there isn’t enough budget to provide services like disaster recovery (DR) using traditional methods and infrastructure. The good news is that public cloud platforms, like AWS, are becoming the de facto infrastructure choice for DR. However, workload portability solutions that simplify cross-platform or cloud recovery are required to meet most RTO & RPO SLAs in the cloud. AWS provides the infrastructure we need to bring DR to tier 2 and tier 3 workloads that have never been able to afford it before. Now, we need orchestration and automation to make it scalable and reliable.
In this session you will learn key considerations and practical steps for getting to the AWS cloud and how you can leverage Amazon S3 storage for cost-effective disaster recovery. Dow Jones will also share details on their migration to AWS Cloud, the benefits realized there, and what the future looks like. Session sponsored by Commvault.
2. Agenda
• What are you looking for in the cloud?
• Key Considerations for DR to the Cloud
• Initial Protection — Data to the Cloud
• Seeding Cloud Instances
• Disaster! – DR Scenario with AWS
• DR Case Study using Commvault and AWS
• Protection in the Cloud – AWS Focused
• Key Takeaways
• Q & A
2
5. Key Considerations for DR to the Cloud
5
• Multiple Sources that are Interrelated. Physical, Virtual, different OSs and
Applications
• Production is Dynamic – The DR Plan for today doesn't work Tomorrow
• Is this a full or Partial DR
• Prioritization – Core, Tier 1, Tier 2 and then Tier 3
• Should Work Versus Knowing it will Work
• Test, Test, Test
• Spot test key Applications – Virtual Application Testing
7. Getting Data to the Cloud
Storage / offsite tape replacement
• AWS and 40+ other storage targets
natively supported
• Combined local backup, archive with cloud
offsite
• Short-term retention for fast local recoveries
• Dedupe, encryption pre-cloud transfer
• Distributed indexes — data movers are
commodity resources, restore from anywhere
• Incremental forever — efficient WAN utilization
• Multiple accounts / account segmentation
supported
• Data portability
7
8. Not All Data Can Be Sent Over the Wire
• Large-scale datasets
(petabytes)
• De-duplication baseline,
or incompressible
datasets
(video, audio)
• Bandwidth $$
• Physical limitations and
reliability
of WAN links
8
Dataset Size
Link
Size
1 GB 10 GB
100
GB
1 TB 10 TB 100 TB 1 PB 10 PB
10 Mbit 14m 2.2 hrs 22.2 hrs 9.2 days 92.6 days – – –
100 Mbit 1m 20s 13m 20s 2.2 hrs 22.2 hrs 9.2 days 92.6 days – –
1 Gbit 8s 1m 20s 13m 20s 2.2 hrs 22.2 hrs 9.2 days 92.6 days –
10 Gbit 0.8s 8s 1m 20s 13m 20s 2.2 hrs 22.2 hrs 9.2 days 92.6 days
9. Achieving Petabyte Scale Data Transport
CLOUD LIBRARY SEEDING SUPPORT
• Not all datasets can be sent over WAN –
physical transportation required for large scale
transfers
• Parallelize as needed
• AWS transfer achieved = Transfer more than 1
petabyte per week
• Simplified Logistics
• End-to-end Custody chain
• De-dupe via Commvault
10. Amazon Storage Class Selection
Ensuring “fit-for-purpose”
• Amazon S3 - Hot
• S3-IA - Warm
• Amazon Glacier - Cold
• Cost models
• Ingress, egress
• Recall thresholds
• Special conditions
• Use cases
• Short- vs. long-term retention
• Backup vs. archive vs. compliance
• Data retrieval policies
(Amazon Glacier)
10
11. Backup to the Cloud
Table stakes – S3, S3 IA, Amazon Glacier
11
15. Replicate and Recover to the Cloud
15
Data Protection Operation is run
against VMs to be replicated
After Protection Operation is Complete
Changed Blocks are Shipped to Standby
VMs
2
OR DASH Copy is Performed to
Secondary Site
Then Changed Blocks are Overlaid on
Replicated VMs
4
Primary Datacenter
Secondary Datacenter
From VMware to AWS
31
43
21
4
16. DR to the Cloud
VMware recovery to AWS — Recovery and Replication
Let’s See it in Action!
16
17. “Now that I’m in the
cloud, I don’t need
backup … right?”
19. Reasons to Back Up in the Cloud
Expect components to fail, design for it – don’t wait for it to happen
Data loss can happen
• Instances retire
• Applications can become corrupted
19
21. What Should I Be Protecting?
Data classification matters — identifying valued data
Stateful machines
• Protect with VSA, BLB, FS, App agents
Stateless machines — “worker” nodes
• Where’s the valued data written out to? Protect that instead.
Data silos
• Object storage, Amazon EFS, Amazon EBS
21
22. Virtual Server Agent for AWS
Expanding Capabilities
22
Agent-less backup of an Amazon EC2
instance’s EBS volumes
Use agent-in-guest for Application
Consistent protection, and as part of a
comprehensive data protection
strategy
23. What’s New? - EBS Snapshot Management (VSA for AWS context)
• Snapshots auto-rotated based on # of Snaps, or Days (Retention)
• Extraction to S3 optimized snapshot storage model to reduce storage size
• Copy Management features (Multiple Copies, cross-account/providers)
• Known snapshot names (SP_2_10242_2266_1468318241) and Tags to differentiate
between Commvault-created and other snapshot
25
24. Amazon RDS Support
Protecting Database-as-a-Service
26
Protect & Manage Databases -- whether
they live in Amazon RDS, on EC2-
instances, or both!
• Data protection for Databases running as a
service within Amazon RDS
• Customer can protect Databases whether
they reside in an EC2 instance, or in Amazon
RDS
Supported mechanisms
• Application-consistent Snapshot backup
Snap Restore, Snap Replicate
25. Keeping It Simple
Use the right tool for the job
VSA for Cloud Image: Agentless instance protection
Agent-in-Guest: App-consistent protection (File, SQL, Oracle, many
more!),
granular app-specific restore features
DBaaS: Snapshot orchestration (coordinate with other systems)
Cloud Snapshot Management: (Oracle on Linux, Linux FS)
27
26. Protection in Cloud
You Need to Protect once you Fail Over
Protection of AWS workloads
Let’s See it in Action!
28
27. Dow Jones / News Corp NA
Shaown Nandi – VP Head of
Infrastructure and Cloud
28.
29. CLOUD TRANSFORMATION JOURNEY
• Save $100 Million annually
• Reduce from 50 data centers globally to 6
• 75% of computational power in the Cloud
AUDACIOUS GOALS SET IN FY14 BY NEWS
CORP
Increase in AWS instances
400%
Dow Jones cloud
compute
58%
30. 32
Dow Jones journey to the cloud with Commvault and
AWS
1. Challenges
2. Preparation
3. Migration
4. Benefits
5. Cloud of the Future
Provision Cloud Resources Automated Recovery
Cloud Orchestration Workflow Automation
31. Dow Jones Legacy DR Operation and Challenges
Traditional DR Strategy Included:
• Management of over 30,000 tapes
in cycle
• Poor RoI
• Offsite Storage Costs
• Unacceptable recovery SLA’s for
mission critical applications
• Limited resiliency
33
Traditional DR Operations
Lowest Cost
Slowest Recovery
User Data Sets
Files, eMail, Mobile, Laptop, etc
32. Preparation & Planning
Leveraged Commvault and AWS Extensive Planning Resources
34
• Jointly Validated Reference Architectures for Commvault and AWS
• Cloud Transformation Services with Commvault
• Commvault Cloud Architecture Guide for AWS
• AWS High Availability Best Practices
Commvault and AWS Jointly Validated Reference Architectures
33. Dow Jones Cloud Based Protection Solution – Real World
Results
Resilient, Automated In Cloud Protection with Commvault
35
1. Started sending on-premise
backups to S3 and S3-IA –
exploring Glacier
2. Ultimately protect over 4,000
instances in the cloud with
Commvault
3. Investigating protected VM’s
with Commvault directly into
EC2
4. Building out full business
continuity plan to include
automated disaster recovery
testing and failover.
36. Commvault Ties together on-prem and cloud data strategies
Commvault Orchestrates the
Enterprise
• Back up in the Cloud
• Back up to the Cloud
• Disaster Recovery to the Cloud
• Workload Portability
38
AWS and Commvault together combine to
minimize networking, storage and
infrastructure costs, while providing the
business a sound data protection and disaster
recovery strategy.
38. Key Takeaways
40
• Consider the Multiple Sources
• Keep the Plan Current
• Know what is going Where
• Bring up the key bits First!
• Test and know it Works
• DR doesn't need to be an Annual Event. Test Applications all year
long.