Continuous Availability Over Distance - Dell EMC · Continuous Availability Over Distance Extending Clusters Across Your ... VMware Continuous Availability w/ VPLEX All inside DCs

1 © Copyright 2013 EMC Corporation. All rights reserved.

Continuous Availability Over Distance Extending Clusters Across Your Distributed Data Center

Muhammad Salama Systems Engineer, Team Lead


Why Continuous Availability?


Why Continuous Availability?


Unplanned Outages: Disaster Strikes!

POWER FAILURE EARTHQUAKE

FIRE FLOODS TECHNICAL

ERRORS

WIND STORM


Planned Outages: Juggling Act!

TECH REFRESH

RELOCATION

WORKLOAD REBALANCING

MAINTENANCE


DOWNTIME COST

$1.8 million per day

$45,000 per hour

$750 per minute

In an “Always On” world, loss of critical applications can be fatal SOURCE

• Business Disruption • Lost Revenue • End User Productivity • IT Staff Time Allocation • Fines and Penalties • Company Reputation

* Enterprise Management Associates - 2012

The Cost of Downtime


http://middle-east.emc.com/microsites/2012/emc-brs-survey/index.htm


http://middle-east.emc.com/microsites/2012/emc-brs-survey/index.htm


Disaster Recovery Tape Backup and Replication

High Availability In-data-center Clusters for Application Protection

New Levels of Business Continuity

Application continues with no disruption (Zero Downtime)

Continuous Data Protection Operational Recovery to Any Point in Time

Continuous Availability Active-Active Data Centers

The Evolution of Business Continuity


A New Business Continuity Strategy FLEXIBLE

Impact

AVAILABLE

Downtime

Continue to Operate

Move Data Non-Disruptively

OPERATIONAL

Recovery

Rollback to ANY Point-in-Time

0 ∞ 0

Continuous Availability Data Mobility Continuous Data Protection


DATA MOBILITY = INFINITE FLEXIBILITY

CONTINUOUS AVAILABILITY = ZERO DOWNTIME

Enter EMC VPLEX


USE CASE #1: Site Failure

Disaster Recovery

Active Site Passive Site



Disaster Recovery

Active Site Passive Site RTO: ~ Hours



Disaster Recovery

Passive Site Active Site RTO: Hours


SITE 2 SITE 1

Continuous Availability During Site Failure

Distributed Virtual Volumes Distributed Virtual Volumes

ESX Stretch Cluster


Continuous Availability During Site Failure


VMware Continuous Availability w/ VPLEX

All inside DCs -> Across DCs – HA (up to 5 ms RTT) – FT (up to 1 ms RTT) – vMotion (up to 10 ms RTT) – DRS (up to 10 ms RTT)

Key deployment details – Stretched layer 2 network – Setup affinity rules for DRS – No requirement for VMware SRM

Data Center 1 Data Center 2

DISTANCE

VMware DRS VMware HA

VMware FT

VMware vMotion

Certified

Distributed Virtual Volume


Oracle Continuous Availability w/ VPLEX

All inside DCs -> Across DCs – RAC (up to 5 ms RTT) – Oracle Live Migration (up to 5 ms

RTT)

Key deployment details – Stretched layer 2 network – No requirement for ASM mirroring

Data Center 1 Data Center 2

DISTANCE

Certified

Oracle OVM Live Migration

Oracle RAC

Distributed Virtual Volume


USE CASE #2: Operational Failure

POWER FAILURE

EARTHQUAKE

FIRE WIND STORM

What must you do? - Lengthy RPO to the last known good backup or snap shot

- Lengthy RTO for coordinating recovery from offline sources

- Complex and time consuming

Is there a better way?

FLOODS

TECHNICAL ERRORS


RecoverPoint for Operational Recovery

Distributed Virtual Volumes

RecoverPoint is: A SAN-based data replication solution which provides operational recovery and 3rd site replication for VPLEX environments

RecoverPoint Provides: Continuous local and remote protection with WAN bandwidth reduction

Journal


WAN/SAN

How it works

How RecoverPoint Works

RPAs (cluster)

Journal

Local Protection

Remote Protection

Host

Journal

RPAs Splitter

Local Copy

r A r C r B

Remote Copy

r A r B


RecoverPoint Benefits

CONTINUOUS

12AM 12AM

BACKUP

Provides DVR-like, ANY Point-In-Time access to data

DATA CORRUPTION OCCURS AT 3:28 pm

12+ Hrs

3+ Hrs

8AM 10AM 12PM 2PM 4PM 6PM 8PM 10PM 12AM

3:27PM

SNAPSHOTS

8AM 11PM 1PM 6PM


Planned Outages: Juggling Act!

TECH REFRESH

RELOCATION

WORKLOAD REBALANCING

MAINTENANCE


A traditional tech refresh can cause months of planning, downtime and will be expensive. With VPLEX, non-disruptive migrations can save time, money and eliminate downtime and risk.

USE CASE #3: Tech Refresh


Timeline

Purchase and Install New

Storage

Migration Planning

Weekend 1 Batch A Outage & Data Copy

Weekend 2 Batch B Outage & Data Copy

Weekend 3 Batch C Outage & Data Copy

Weekend 4 Batch D Outage & Data

Copy

Completed Storage Refresh

Migration Time

Traditional Tech Refreshes

Non-reversibility Wastes employee time Long time to value

Expensive professional services engagement Disruptive to applications Remediation required


LUN2

LUN1

Continuous Availability During Tech Refresh

1. New array is brought in

2. VPLEX discovers new array and devices

3. VPLEX mirrors storage

4. I/O is automatically and non-disruptively directed to the new array

5. Once the new array is fully replicated the old array is removed automatically and without disruption

6. The old device can be removed from VPLEX and rolled out of the data center

LUN2

LUN1


New Migration Strategy with VPLEX

Timeline

Purchase and Install New

Storage

Migration Planning

Weekend 1 Batch A Outage & Data Copy

Weekend 2 Batch B Outage & Data Copy

Weekend 3 Batch C Outage & Data Copy

Weekend 4 Batch D Outage & Data

Copy

Migration Time

Completed Storage Refresh

Non-Disruptive DATA MOBILITY

During Working Hours

Migration Time

Save money Non-disruptive migrations FOREVER No remediation

Eliminate risk Save employee time Faster Time to Value


USE CASE #4: Data Center Relocation

Significant Planning High Cost High Risk Highly Disruptive


SITE 2 SITE 1

Continuous Availability During Data Center Relocation

Distributed Virtual Volumes Distributed Virtual Volumes


SITE 2 SITE 1



VMware vMotion, HA, FT



Avoid the potential risk, cost and disruption of a data center relocation.

Lead the way with continuous availability; fast, efficient and safe


Traditional load balancing is • Complex to plan • Slow to implement • Disruptive to applications Leverage distributed virtual volumes to immediately balance loads across arrays.

USE CASE #5: Performance Load Balancing


When your array is 100% utilized, traditional moves require downtime.

0%

100% I/O load

50%

App1 The applications relying on the data are shut down to relieve I/O pressure.

50% I/O

0%

100% I/O load

50%

50% I/O

App2 App2

Traditional Performance Load Balancing

LUN1 LUN2

Data is replicated to new array • Hosts and array are remediated • Hosts and paths are reconfigured • Hosts are rebooted

Access is restored Assuming the changes didn’t break something else!


0%

100% I/O load

50%

App1

50% I/O

0%

100% I/O load

50%

50% I/O

App2 App2

Load Balancing with Continuous Availability


As hosts consume array resources such as space and I/O… VPLEX can non-disruptively relocate work loads Applications access to distributed virtual volumes never changes

• No remediation • No reconfiguration • No reboots • No downtime • And confidence in changes

LUN2

LUN1


EMC VPLEX for Continuous Availability

Flexibility

Availability

Mobility


Helping Organizations Make the Journey

EMC Continuous Availability Assessment Service

• Current-State Snapshot

• Application Readiness Assessment

• Continuous Availability Business Case

• Implementation Roadmap & Timeline


Continuous Availability Over Distance - Dell EMC · Continuous Availability Over Distance Extending Clusters Across Your ... VMware Continuous Availability w/ VPLEX All inside DCs

Documents