The Importance of High Availability

Fault tolerance is no longer a 'wish list' item, but a 'must-have'. It is no longer acceptable for services to suffer outages or even to be slow to respond during peak traffic periods.

It is therefore of paramount importance to ensure that your services keep running, and can cope with traffic load, at all times. ZXTM allows you to set up fault-tolerant clusters in highly-flexible configurations of any number without the constraint of having to pair up units.

Quick Links

I've already got a pair of load balancers, isn't that enough?
What can ZXTM do for me to make service fail-over more flexible?
How is this flexible high availability going to help me?
What can ZXTM do to help me enforce my differentiated SLAs?
Can you show me examples of typical high-availability deployments of ZXTM?

I've already got a pair of load balancers, isn't that enough?

Most current high-availability solutions provide a system that if one critical component fails, a single extra machine is available for redundancy, but the system is now very vulnerable to a subsequent compound failure. This is known as N+1 redundancy. A typical example of this configuration is a fault-tolerant pair of front-ends. Whether in a active/active or a active/standby configuration, a failure in one front-end leaves you with a single point of failure for your business-critical application traffic until the failed machine is restored, effectively 1+1 redundancy.

For increasing numbers of companies, this high-availability configuration is no longer satisfactory. Instead, multiple highly-available layers are required within the traffic management layer, so that even compound failures can be guarded against. The trend is towards traffic management 'clouds' where any of the machines within a particular cloud will ensure continuity of service despite compound failures. This is known as TrafficCluster™.

The traffic manager cloud model blurs the traditional distinction between 'master' and 'slave', so that any combination of active and standby front-ends is now possible. This ensures that your traffic management framework can be scaled horizontally to provide many high-availability layers as your business requires.

What can ZXTM do for me to make service fail-over more flexible?

It is not only at the front-ends that flexible high availability is required; the ability to provide fully customisable fail-over behavior for the back-ends applications is also paramount. Flexibility is required because:

applications are often not designed to be cluster-aware; and
different types of application have differing failure symptoms and recovery actions.

ZXTM provides you with completely flexible high-availability configurations for both your traffic managers and your back-end services. Every aspect of the fail-over behavior can be customized using our powerful, yet easy to use TrafficScript™ language. For your convenience, ZXTM also provides you with graphical configuration wizards to configure common fail-over functionality and write TrafficScript for you.

How is this flexible high availability going to help me?

You will need flexible high availability whenever you need to specify in your traffic manager what counts as a failure in your back-end services and what actions should be taken to remedy that failure. For instance, a database may only be classified as 'failed' on the basis of certain types of SQL query; the recovery action may require a new database to be reconfigured at the application level in order for it to take its place. It is a certainty that these actions will vary greatly between differing SQL databases such as Microsoft SQLServer, Oracle 9i, MySQL and Postgres.

ZXTM can designate a back-end server as 'failed' on the basis of generic tests such as ping or connection to the published service. In addition, more complex tests can be written by you or by Zeus to determine what additional criteria signify that a service has failed. This ensures that both application and hardware failures can be dealt with automatically, maintaining the quality of your users' experience.

What can ZXTM do to help me enforce my differentiated SLAs?

Detailed information on how ZXTM helps you monitor and enforce your Service Level Agreements is available under Traffic Management.

Can you show me examples of typical high-availability deployments of ZXTM?

Many real-world high-availability deployment solutions can be found in our Solutions sections. If you would like to discuss which solution would best fit your needs, please call us on +44 (0)1223 568555 or drop us an email on sales@zeus.com.