Database and Public Endpoints redundancy on Azure

  • View
    273

  • Download
    2

  • Category

    Science

Preview:

DESCRIPTION

Database and Public Endpoints redundancy on Azure Radu Vunvulea

Citation preview

Radu Vunvulea@RaduVunvuleahttp://vunvulearadu.blogspot.com

DATABASE AND PUBLIC ENDPOINTS REDUNDANCY ON AZURE

OUR SPONSORS MADE THIS POSSIBLE!THANK YOU!

Click icon to add picture

Transylvania, Romania

RADU VUNVULEA MCTS MCP BANK HOME AUTOMATION

MVP ENTERPRISE

AUTOMOTIVE PHARMA

LEAN AND AGILE E-COMMERCE

WEB iQuest

AZURE JAVASCRIPT VUNVULEARADU.BLOGSPOT.COM

MOBILE DOTNET @RaduVunvulea

WCF WPF ENTHUSIASTIC

CLOUD IS NOT MANAGED BY GOD

AGENDA

• Define Redundancy • SQL Database• Storage• Endpoints (HTTP/S, WCF)• VM

In engineering, redundancy is the duplication of critical components or functions of a system with the intention of increasing reliability of the system, usually in the form of a backup or fail-safe.

Source: Wikipedia

REDUNDANCY

HardwareDual Modular RedundancyTriple Modular RedundancyInformationError detectionCorrectionTime Transient Fault DetectionSoftwareN-versions Programming

TYPE OF REDUNDANCY

HardwareDual Modular RedundancyTriple Modular RedundancyInformationError detectionCorrectionTime Transient Fault DetectionSoftwareN-versions Programming

TYPE OF REDUNDANCY

HardwareDual Modular RedundancyTriple Modular RedundancyInformationError detectionCorrectionTime Transient Fault DetectionSoftwareN-versions Programming

TYPE OF REDUNDANCY

HardwareDual Modular RedundancyTriple Modular RedundancyInformationError detectionCorrectionTime Transient Fault DetectionSoftwareN-versions Programming

TYPE OF REDUNDANCY

HardwareDual Modular RedundancyTriple Modular RedundancyInformationError detectionCorrectionTime Transient Fault DetectionSoftwareN-versions Programming

TYPE OF REDUNDANCY

HOW WE CAN TEST IT?

Netflix: The Chaos Monkey’s job is to randomly kill instances and services within our architecture

CHAOS MONKEY

CHAOS MONKEY

?FAILOVER & SWITCHOVER

FAILOVER & SWITCHOVER

SQL REDUDANCY

DEMO

LET’S SEE SOME CODE

SQL BACKUPS

DEMO

LET’S SEE SOME CODE

ALWAYSON

DISASTER RECOVERY PLAN

Do you have a disaster recovery plan?How long it takes to restart the entire system?How complicated is your checklist?Can you parallelize it?

DISASTER RECOVERY PLAN (DRP)

Do you have a disaster recovery plan?How long it takes to restart the entire system?How complicated is your checklist?Can you parallelize it?

AUTOMATION

DISASTER RECOVERY PLAN (DRP)

ENDPOINT REDUNDANCY

• Any public endpoint• Accessible from internet or private networks

ENDPOINT

LOAD BALANCER

• Free • Out of the box• Lever 3 Load Balancer type

– TCP Level• No Sticky sessions• Partial Configurable

LOAD BALANCER

• Custom app endpoint• Processor level, 500 Errors, Parallel task • Configurable from *.cscfg files

• VM: Using Power Shell

LOAD BALANCER PROBE

<LoadBalancerProbes> <LoadBalancerProbe name="FooProbe" protocol="http" intervalInSeconds="30" path="/api/probe" port="80" timeoutInSeconds="60" /> </LoadBalancerProbes>

REDUNDANCY

TRAFFIC MANAGER

TRAFFIC MANAGER

TRAFFIC MANAGER

• ~Load Balancer (Word Wide level)• Redirect request based on client location• Any kind of endpoint

– VM, Web/Worker Roles, DB and many more– External Endpoints

• Methods: Failover, Round Robin, Performance• Is not FREE (1B DNS query = $0.75)• DNS TTL

DEMO

LET’S SEE SOME CODE

TRAFFIC MANAGER

• [Not Supported] Hybrid Scenarios

WHY REDUNDANCY IS SO IMPORTANT?

• Compute 99.95%• SQL Database 99.90%• Storage 99.90%• Storage (RA-GRS) 99.99%• Active Directory 99.90%• Cache 99.90%• VM 99.95%• Service Bus 99.95%

WINDOWS AZURE SLA’S

• Compute 99.95% 21.6m• SQL Database 99.90% 43.2m• Storage 99.90% 43.2m• Storage (RA-GRS) 99.99% <8m• Active Directory 99.90% 43.2m• Cache 99.90% 43.2m• VM 99.95% 21.6m• Service Bus 99.95% 21.6m

WINDOWS AZURE SLA’S (PER MONTH)

• Compute 99.95% 21.6m• SQL Database 99.90% 43.2m• Storage 99.90% 43.2m• Storage (RA-GRS) 99.99% <8m• Active Directory 99.90% 43.2m• Cache 99.90% 43.2m• VM 99.95% 21.6m• Service Bus 99.95% 21.6m

WINDOWS AZURE SLA’S

• Downtime per month: 216m21.6*2+43.2*4=216m

WINDOWS AZURE SLA’S

• Downtime per month: 216m21.6*2+43.2*4=216m

• Downtime per month: 0.5%(100/43200)*216=0.5%

WINDOWS AZURE SLA’S

• Downtime per month: 216m21.6*2+43.2*4=216m

• Downtime per month: 0.5%(100/43200)*216=0.5%

• Uptime: 99.5%100-0.5=99.5%

WINDOWS AZURE SLA’S

• Compute 99.95% 21.6m• SQL Database 99.90% 43.2m• Storage 99.90% 43.2m• Storage (RA-GRS) 99.99% <8m• Active Directory 99.90% 43.2m• Cache 99.90% 43.2m• VM 99.95% 21.6m• Service Bus 99.95% 21.6m

WINDOWS AZURE SLA’S

WCF ENDPOINTS REDUNDANCY

SERVICE BUS RELAY

Relay

• Steps to integrate in existing application– Add assembly reference to SB Relay– Update WCF configuration sections

(client and backend)

SERVICE BUS RELAY

• Works only with– Async calls– Fire and Forget calls

• Broker between endpoint and client• Store messages until backend is available• WCF Endpoint (Azure or on-premises)• Price model similar with SB (per messages)

SERVICE BUS RELAY

STORAGE REDUNDANCY

• 3 copy of the content in the same data center

LOCALLY REDUNDANT STORAGE

1

3

2

• 2 different data centers• Async replication

GEOGRAPHICALLY REDUNDANT STORAGE

1

3

2

1

3

2

• Similar with GRS• Read only access to the secondary storage• Redirecting managed by developer• Async replication• 99.99% Uptime

READ ACCESS GEO REDUNDANT STORAGE

DEMO

LET’S SEE SOME CODE

TRAFFIC MANAGER

TRAFFIC MANAGER

CHECK HEALTH STATUS OFOTHER SERVICES

IMPORT EXPORT SERVICE

IMPORT/EXPORT SERVICE 

DEMO

LET’S SEE SOME CODE

MTBF, MTTR

Mean Time Between FailuresMean Time To Recovery

MTBF -> MTTR

VM REDUNDANCY

• Same features offered by System Center on-premises

• Availability Set• Windows Server Failover Cluster

– Real time Sync

VM

CONCLUSION

• Everything has a price• Calculate up-time for specific roles and use-case• Don’t trust the 99.99% myth• Be as flexible as possible• Automate • Recovery plans

CONCLUSION

Q & A

OUR SPONSORS MADE THIS POSSIBLE!THANK YOU!

Recommended