The cloud scores over on-premises computing mainly owing to the resilience and redundancy it offers. Cloud providers maintain data in multiple servers located at different parts of the world so that even if one data center is down for any reason, another one can take its place and ensure normal service. At least, this is how it is supposed to work in theory.
The reality can be different as Amazon Cloud Services has repeatedly found out the hard way.
In 2011, a lightning struck Dublin, destroying a power transformer and taking down the power supply to Amazon’s data center along with it. That by itself was nothing extraordinary, as any data center worth its salt have power backups in place. The problem was that the same lightning that destroyed the power backup infrastructure at the data center plunged Amazon’s EC2 cloud service offline as well. The much-touted redundant mechanisms characteristic of cloud computing came to naught in this case!
In June 2012, Amazon’s Northern Virginia data center suffered from similar power outages twice in less than a month. This outage, caused by electric storms, even affected services such as Netflix, Instagram and Pinterest. These incidents drive home the bitter truth that the cloud, for all its advantages, is by no means infallible.
Cloud providers would do well to learn from the disaster that has repeatedly stuck Amazon to pay some serious attention to the recovery and restoration processes they offer.
A backup may not exactly be a backup after all, especially when it is interconnected to the main source. Just as storing data in a CD disk will not help if fire engulfs the premises and burns down both the hard disk and the CD storage facility, cloud providers need to properly diversify when it comes to security. In Amazon’s power outage, backup power could ideally have been from a different grid.
Customers would likewise do well to have their own redundancy and disaster recovery systems in place rather than blindly trusting the cloud provider. Listed below are a few things the customers should consider on their part:
- Master the redundancy and fail over plans available with the cloud service providers. Amazon’s Availability Zones, for instance, allow the customers to set up cloud-based redundancy on their own.
- Look for visibility over and above reliability. The customer needs to understand what exactly is happening with the data entrusted with the cloud service provider. Make sure that the service level agreements with the cloud service provider have teeth.
- Sign up with more than one cloud service provider which would take resilience and redundancy to a deeper level. Side by side, there is a need to ensure interoperability between the different cloud platforms.
- Have a strong disaster recovery plan in place, fully incorporating the realization that even the cloud may face serious downtimes.
- Cloud computing may in fact be just as or even more vulnerable than on-premises computing. Only the nature and dimensions of threats may differ. It requires a proactive effort to ensure that the data stored in the cloud remains safe and secure.