Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Paul Ignatius & Warren Perlman
PB03374PUS
#VMWorld #PB03374PUS
PanelvCloud Availability –Real World Architecture Details on VMware’s Latest Replication Tool
VMworld 2017 Content: Not fo
r publication or distri
bution
• This presentation may contain product features that are currently under development.
• This overview of new technology represents no commitment from VMware to deliver these features in any generally available product.
• Features are subject to change, and must not be included in contracts, purchase orders, or sales agreements of any kind.
• Technical feasibility and market demand will affect final delivery.
• Pricing and packaging for any new technologies or features discussed or presented have not been determined.
Disclaimer
2#PBO3374PUS CONFIDENTIAL
VMworld 2017 Content: Not fo
r publication or distri
bution
Agenda
#PBO3374PUS CONFIDENTIAL 3
1Introductions
• Background
2 Replication concepts
3An Enterprise’s journey to availability, problems
and architecture
4 vCloud Availability Overview
5vCloud Availability & VMware Cloud Director
implementation at Navisite
6 Key takeaways for successful implementations
VMworld 2017 Content: Not fo
r publication or distri
bution
Paul Ignatius
CTO, Navisite
@storageNerd
Warren Perlman
CIO, Ceridian
vCloud Availability – Real World Architecture Details on VMware’s Latest Replication Tool
4#PBO3374PUS CONFIDENTIAL
VMworld 2017 Content: Not fo
r publication or distri
bution
Hard Truths about Replication…
• “Replication recovery in most cases are crash consistent”
• If you recover an application built on top of multiple servers, you are recovering and praying
• In the past replication mostly supported DR. Availability is a post 2004 phenomenon
• Availability has needs above and beyond basic replication
• Replication copies are used to source backups, stand by copies for test, reporting assets
• Virtualization and cloud based technologies enabled reduction in needs of identical hardware and allowed smaller compute footprint and on demand compute consumption during fail over
• 95% of recovery and fail over is to demonstrate you can when that one actual recovery occurs.
• Application resiliency: is what most people look for recovery solutions for
– In the cloud age, this is table stakes, big boys play with contractual resiliency
• Contractual resiliency: portability of an application across service providers with ability to demonstrate service resiliency is now required by law.
#PBO3374PUS CONFIDENTIAL 5
VMworld 2017 Content: Not fo
r publication or distri
bution
Evolution of Replication Based Technologies
• Change Capture and Copy on Write (file system and volume level) technologies have enabled
– Snapshots, Replication, Backup
• At the core of it, a driver grabs the file block or volume block being changed, either write it in place, copy out old data, or grab the new copy, replicate it else where, apply to a change log
– Create point in time volume images based on snapshots
– Allow all changes to be logged so that various continuous snapshots can be made available
– Over a period of time consumes large amount of data
– Drove consolidation to known recovery intervals so that the replicas became based on copy on write
– Crash consistent point in times also mean lesser data on the replica
– Integrate with application to create consistent images at various book marks
– Consolidate the data @ definite intervals and also create full images based on performance needs (RTO)
• Seeding replicas simplify initial kick off
6#PBO3374PUS CONFIDENTIAL
VMworld 2017 Content: Not fo
r publication or distri
bution
Applications of Replication Based Technologies
• DR
• Snapshot of replicas (point in time recovery)
– Good for creating backups
– Reporting engine data feed
– Test server data feed
• Copy on write images on the replica
– Ability to mount snapshot of the replica while replication is continuing
– Verify recovery data consistency
– Great for testing use cases and demonstrating compliance
• Holy grail – High availability – on demand or upon automatic monitoring, create a writable snapshot of the replica, mount/boot, route traffic to newly mounted infrastructure, test, and promote the replica or destroy
7#PBO3374PUS CONFIDENTIAL
VMworld 2017 Content: Not fo
r publication or distri
bution
Truth about DR & HA
• Replication based recovery/HA actions at large
– <1% are actually about recovering from a natural disaster
– <5% are to recover from a user or application error
– >95% are about testing & demonstrating conformance to compliance controls
• By default systems are crash consistent at DR image and or HA image.
• It’s the application!
– At times, applications span multiple VMs
– Consistent bookmarks/checkpoints/rewind marks are key in recovery
– Should be created across the constituent nodes as prescribed by the application.
8#PBO3374PUS CONFIDENTIAL
VMworld 2017 Content: Not fo
r publication or distri
bution
The Many Sides of Replication – Key Points You Need to Know
Then
• ‘Must have’s for a replication solution in the old days
– Identical hardware and data center configuration
– Special networking gear
– What’s my encryption?, Special hardware
– Fidelity
– Live systems on both ends
– Application resiliency
– File system based
– Recovery
– Speed
– Rewind
– As you may remember
• Doubletake, XOsoft, Neverfail, Inmage, Sunguard….
Now
• Must haves for the 2017
– Simplified deployment
– No special hardware or VPN
– ‘Test while production is protected’
– Born in the virtual world – No CPU/RAM/Just Disk
– Contractual resiliency
– Replica VMs are mostly disk images spun up on demand
• No active consumption of CPU and RAM
– Orchestrated for consistency
• Across Virtual machines
• Application aware
– Secure & Integrated with identity providers
– Recovery testing to demonstrate control compliance
– DR testing as a service
#PBO3374PUS CONFIDENTIAL 9
VMworld 2017 Content: Not fo
r publication or distri
bution
Ceridian – HCM Solutions
10#PBO3374PUS CONFIDENTIAL
VMworld 2017 Content: Not fo
r publication or distri
bution
Ceridian’s Data Protection and Availability Journey
• Explored multiple different DR solution over years
– Colo based DR program - Backup to tape – recovery from tapes
– Disk based backup – data availability at both primary and DR data centers
– Array based replication
– OS agent based replication
– Application based replication with SQL Always-On availability sets
• Now
– Hypervisor level replication between site to site
• Future:
– vCloud Availability and/or replication to the cloud
11
Past, Present and Future
#PBO3374PUS CONFIDENTIAL
VMworld 2017 Content: Not fo
r publication or distri
bution
Ceridian’s High-level Architecture
• Challenges • Challenges
– Multiple Technologies
– Substantial Infrastructure Investments
– RTO of 2 hours
#PBO3374PUS CONFIDENTIAL 12
VMworld 2017 Content: Not fo
r publication or distri
bution
Why vCloud Availability?
• Born in the virtual Era – Just powered off VMs on the other side
• Integrated into vCenter/vSphere
• Application aware by integration with VRO– works across multiple VMs
• Built in testing
• Cost effectiveness
• Data Center and regionally deployed applications work identically
– NO VPNs – natively encrypted data replication
• Recovery system definition is pre-built
#PBO3374PUS CONFIDENTIAL 13
VMworld 2017 Content: Not fo
r publication or distri
bution
Navisite’s vCD/vSphere/vAPP Architecture
vCloud Director Customized
Portal
Client Environments
Navisite Data Center
fiber connectivity
Using NSX+ACI
(optional)
© VMware Inc
#PBO3374PUS CONFIDENTIAL 14
VMworld 2017 Content: Not fo
r publication or distri
bution
Considerations Before You Choose vCloud Availability
• Does vCloud Availability make sense for you?
• Checklist for client readiness
• application constructs (vAPP) portability – are you ready ?
• Do you understand your needs of Recovery point / recovery time objectives?
• Sizing guide
• Your data seeding strategy
• Needs for your recovery testing
• Latency ? Not the same as the pipe ?
• Day 2 Questions
• Provisioning failed-over instances mapped to your security domain
• Test on cadence(demo compliance)
• Seamless connectivity post failover
• Maintenance
– ESXi
– Storage
– VCD
• Scalability and performance guidelines
#PBO3374PUS CONFIDENTIAL 15
Management/Monitoring
Security
VMworld 2017 Content: Not fo
r publication or distri
bution
vCAV, vSphere and vCloud Director in Action
16#PBO3374PUS CONFIDENTIAL
VMworld 2017 Content: Not fo
r publication or distri
bution
Takeaways
• Simplicity – Easy to deploy, no redundant hardware
• Effective alignment of RPO, RTO
• Quick to realize availability – minimal initial replication time
• Seamless application availability in the event of an outage
• Infrastructure cost reduction – its about the $$
• Regionally deployed application resiliency
• Foundational components of contractual resiliency
• Come see us @expo hall, demo, sign up on the landing page
vCloud Availability solves critical problems
#PBO3374PUS CONFIDENTIAL 17
VMworld 2017 Content: Not fo
r publication or distri
bution
Calls to Action
• Get a live demo of vCloud Availability
– Navisite booth #212
• Signup for a free trial
– http://go.navisite.com/30-day-vcloud-free-trial.html
• Learn more about replication to the cloud
– http://www.navisite.com/solutions/managed-data-protection
#PBO3374PUS CONFIDENTIAL 18
VMworld 2017 Content: Not fo
r publication or distri
bution
Paul IgnatiusCTO, Navisite@storageNerd
Warren PerlmanCIO, Ceridian@wperlman
Thank You
VMworld 2017 Content: Not fo
r publication or distri
bution
VMworld 2017 Content: Not fo
r publication or distri
bution
VMworld 2017 Content: Not fo
r publication or distri
bution