What is High Availability

What is High Availability?

When do you need it?

How do you create it ?

How do you measure it?







Lee Hampton-Whitehead

What is High Availability?



“High availability is a system design

protocol and associated

implementation that ensures a certain

absolute degree of operational

continuity during a given

measurement period.”

Source http://en.wikipedia.org/wiki/High_availability

What is High Availability?



• Minimum service / application outages

• Uptime is not the same as availability

• Planned outage considerations

• Quality of Service

When do you need it?



Any system failure that may impact business

function is a candidate for a HA solution

• Internal Own Use systems (Email or Resource

management)

• Customer Advisor (Helpdesk or CRM)

• External Customer Facing (Online Shopping or

banking)

When do you need it?

The cost of failure

• Financial

– Loss of orders

– Share Price

– Contractual penalty clauses

• Brand Value

• Customer satisfaction and loyalty

Other considerations

• Regulatory requirements

• Safety Implications

When do you need it?



Business impact analysis is required to assess

• The cost of service outage

• Total Cost of Ownership (TCO)

• Return on Investment (ROI)

• Budget spend

When do you need it?



Risk analysis factors

• Time to recovery

• Data impact of outage or recovery

• Associated hardware software and resource costs

• Probability of event

• Implications on the business

• Complexity risks

• Third-party solutions

• Pros &Cons

Design Sizing Criteria



• Number of transactions

• Size of transactions

• Secondary load considerations

Service Level Agreement



• Agreed availability criteria

• Measurement techniques

• Penalty clauses

• Capacity, Performance and usage

• Help desk support

• Contingency

• Costing

Costs



The cost implications of most availability solutions

include, but are not limited to, the following:

• Hardware

• Software

• Network infrastructure

• Training

• Serviceability

• Operational costs

Costs

How do you create it ?

Redundant Infrastructure

Effective removal of all SPOF (Single Point Of Failure)

• Power

• Network (including LAN, WAN, DNS etc)

• Server hardware

• Storage and associated interconnect

• Air conditioning

How do you create it ?

Web

Servers









Corporate

Network

Load

Users Balancers



SQL

Servers

How do you create it ?

Load Balancing

• Typically used for WEB based applications

• User requests are distributed across a farm of

servers

• The Load Balancing technology will

redistribute the application traffic in the event

of a server failure

• All servers / hardware actively participate in

delivering the application

How do you create it ?

Failover Cluster

• Typically used for the database components of

an application

• Can be used for File Serving, Print Serving and

Storage solutions where a load balanced may

not be suitable

• A Typical failover cluster deployment (Active-

Passive) may result in under utilised hardware

How do you create it ?

Data Replication / Mirroring

Data is central to any business application

• Log Shipping

• Transactional replication

• Database mirroring

How do you create it ?

Corporate Infrastructure

Basic Requirements

• IP Routed Network

• DHCP

• DNS

• AD

How do you create it ?

Proactive System Monitoring

• Indentify and correct minor / non service

affecting problems before they become major

outages

• Baseline Server utilisation values

• Application / Service monitoring should be

representative of typical user activities

• HA Testing

How do you create it ?

Application Architecture

• Poorly written code can destabilise any HA

implementation

• The Application should be able cope with the

loss of transitory data

• Data Consistency , Use of “Transaction”

How do you create it ?

IT Processes

• Restricted / managed system access

• Software development lifecycle

• Infrastructure life cycle

• Effective service wrap

How do you create it ?

Unapproved changes

Availability Comparison



• 95%

36.5 hours/month or 18.25 days/year



• 99.9% (“Three nines”)

43.8 minutes/month or 8.76 hours/year



• 99.99% (“Four nines”)

4.38 minutes/month or 52.6 minutes/year



• 99.999% (“Five nines")

26.2 seconds/month or 5.26 minutes/year

How do you measure it?

SLA

The Service level agreement will define a

good, poor or failed service in terms

• Service times

• Response times

• Fault resolution times



The SLA will also define how these values are

to be measured

How do you measure it?

Measurement considerations

• Users Perspective

• Monitor every component

• Synthetic transactions

• Simplicity

Availability Report

Summary



• Indentify Business requirements

• Risk Assessment

• Decide on ‘appropriate’ solution

• Indentify and address any SPOF

• Application design is important to ensure high

availability

• Implement and Follow Procedure

• Consider the user story

• Rigorous testing

Q&A