By Susan J. Campbell

Businesses throughout the world rely on their networks to support business processes, run applications and drive revenue for long-term sustainability. As a result, the importance of the network is amplified. The network and the data it maintains must be redundant to ensure optimization in the event of a failure. For this, Alcatel-Lucent recommends the benefits of georedundancy.

A recent article in Alcatel-Lucent’s TechZine by Randee Adams, Consulting Member of Technical Staff, Eric Bauer, Reliability Engineering Manager, and Daniel Eustace, Distinguished Member of Technical Staff entitled, “Availability Benefits of Georedundancy,” explores the enhanced service availability and client-initiated recovery possible with geographically redundant systems. With breakthrough mathematical modeling, a deeper understanding of the availability boost provided by such systems is enabled, while also identifying opportunities for optimization. And, lest we forget, the business continuity assurance it provides in the wake of a major disaster event.

Alcatel-Lucent developed the mathematical modeling that demonstrates how businesses have an opportunity to leverage the existing georedundancy within their systems to enhance overall service availability. This approach also identified potential methods to use for optimization, including retry parameters, faster time out, rapid context restoration and client-initiated recovery.

Beyond the benefits of georedundancy lies its primary purpose – to ensure the continuity of a business following a disaster such as a fire, flood, human error or other impactful event that can render the entire data center or multiple systems useless within one location. While safety of the people is the primary concern, the close second is the recovery of critical services to sustain the business. With georedundancy in combination with methodical disaster recovery plans and manual procedures, service can be recovered in days or even hours.

When the benefits of georedundancy are combined with high-availability mechanisms and systems, service is automatically detected and recovered with a single failure event occurs. High-availability systems must be recovered within seconds, with a benchmark of no more than 315 seconds often set per system per year.

Georedundant systems speed up the time to detect a failure and recover. To enable the benefits of georedundancy, Alcatel-Lucent relies on the continuous time Markov availability models for the quantitative comparison of service availability. Client-initiated recovery is the only proven scheme to address uncovered failures and ensure a fix before the failure is felt by the client’s customers.

One of the main benefits of georedundancy is the ability to improve service availability in the wake of any interruption in service performance. Alcatel-Lucent offers a few recommendations to ensure standard high-availability mechanisms are augmented through georedunant systems:

Client-initiated recovery – use return codes and expiration of time outs to detect critical failures and support recovery to redundant system instances.

Time out and retry parameter optimization – shorten failure detection latency and eliminate false positives.

Enable rigid reestablishment – keep the process to identify, authenticate and authorize a session to another server instance as brief as possible to add only minimal latency.

Rapid context restoration – this must be enabled to retain user data in a common and replicated data store accessible to redundant system instances.

Overload control – when this is implemented, floods of primary and retired client-initiated recovery requests are prevented.

With innovative mathematical modeling provided by Alcatel-Lucent, we now have a deeper view and understanding of the availability benefits of georedundant systems to support a cost-effective, methodical approach to optimization. Reality is that the mathematics as well as historical evidence show that it is always “better to be safe than sorry,” and that when it comes to the protection of a network physical redundancy based on geographic diversification is the best way to ensure network resiliency and very short meantime-to-restoration periods in the face of network outages resulting from disasters.