Natural disasters – such as earthquakes, floods, tsunamis, volcanoes, hurricanes and blizzards – are liable to wreak havoc upon data center operations, causing service and network outages. Although typically unexpected, the question is unfortunately when rather than if such extraordinary events will occur. Still, their effects can be contained by preparing for the inevitable aftermath.
“Hope for the best - expect the worst”
With the possibility of natural disasters and weather-related strikes at any time, there is the possibility that these events produce tremendous consequences to major companies, including the loss of power, power grids shutting down, and massive effects on mass transportation. A prime example is the volcanic eruption in Iceland from earlier this year and its huge effect not only in that region, but all over Europe -- and ultimately on a global scale. Eyjafjallajokull's 7-mile-high plume of volcanic gases and silicate ash spread across much of Europe, bringing air travel across the continent to a near standstill. Although aviation is an obvious casualty of many weather-related disasters, much can be learned from a network and applications viewpoint; specifically, what precautions should be taken to avert data center disaster.
The Need for a Disaster Recovery Solution
A natural disaster obviously impacts ‘ground zero’ – the immediate proximity of the event. During many weather catastrophes, people become stranded/get stuck in massive traffic jams, have limited access to facilities and in some cases shut businesses down entirely. This means that online services and applications – just like their brick & mortar counterparts – become unavailable; whether the cause is lack of personnel to operate these applications and their supporting-systems, or other physical or technical issues caused by the disaster. Such situations are referred to broadly as data center disaster, meaning that the data center - including all of its applications, services and infrastructure - are interrupted and become non-operational.
To overcome a data center disaster, it is recommended to design and implement a disaster recovery (DR) solution – significantly earlier than actually waiting for to occur. This requires planning and deploying a secondary data center – in addition to the original, “primary” one – which runs the same applications concurrently and delivers the same services under the original service level agreement (SLA). Rolling out such a project implies covering all of a data center’s operational aspects, including storage, networking, servers, electricity, cooling and more. Far from a simple task, this involves taking into consideration capacity demands and the projected load at the time of a disaster, which is a feasible and realistic practice. Once in place, implementing a global server load balancing solution in both data centers, ensures that all traffic and user transactions are transparently redirected to the secondary data center – so that for the end-user all applications and services are up and running – unaffected from the disaster.
The Plot Thickens: The Disaster Effect on Additional, Indirect Services
But natural disasters not only affect physical surroundings - the affects can spread for miles. For instance, when an airport in one city closes, the rerouting all of its departures and arrivals to other airports is necessary – causing a significant increase in transactions and processing in the computing systems at neighboring airports. Similarly, the closing of dozens of airports across the U.S. could create a massive traffic surge not only on airport systems – but to many other online transportation services, including railway websites, car rental services, and other local Internet-based services.
This traffic “spike” means unexpected user demand on Internet websites and massive load on data center servers and infrastructure – sometimes up to 50x more than the typical, “baseline” traffic rate. To address this abrupt traffic growth, network infrastructures must be able to process such an increase in capacity. This means that it should not only be highly-performing so that it can process the increased network traffic, but it also needs to be able to scale its capacity “on demand”. An on demand infrastructure provides the flexibility of adding more throughput capacity and services with no hardware replacements or causing any business downtime. Leveraging this on demand approach not only reduces the total cost of ownership (TCO) and increases return on investment (ROI) – but also ensures addressing changing performance capacity needs successfully – even at the harshest of times - which typically arise due to unexpected natural disasters.