Spring may be just around the corner, but extreme weather conditions, including intense cold, can hit at any time, depending on your location. Extreme cold can bring challenges with cloud connectivity, and while data centers are usually not affected by the cold, if outdoor equipment frosts over, it can be difficult for air to circulate, which can potentially shut down the entire system. Because speed and reliability are critical for organizations hosting their infrastructure in a public cloud, the cold could negatively impact thousands of businesses. With this in mind, how can businesses be sure that their public clouds will remain connected during the cold and develop a disaster recovery plan with their cloud strategy in mind?
We had the opportunity to speak with Grant Kirkwood, Co-Founder and Chief Technology Officer of cloud computing solution provider, Unitas Global, about how companies can best maintain connectivity during extreme cold. With seven years of experience at Unitas global, Kirkwood was able to provide us with insight on the subject.
How can companies ensure they maintain public cloud connectivity through the extreme cold?
To ensure public clouds maintain connectivity through extreme weather, companies must maintain diversity in their connectivity architecture through not only carrier diversity, but also physical diversity.
For example, if you’re in a tall high rise in a big city, like Chicago, somewhere in the basement is a room with conduits coming in from below ground. This building may have four to five carriers providing connectivity. Many times, these carriers piggyback on top of each other in terms of the physical entrance, meaning one carrier may buy rights on a different carrier’s entry, similar to how Amtrak pays to run their trains over freight train tracks they don’t own. Understanding that physical entry into the building and diversity on that level is important. If you’re getting cloud connectivity from two carriers that come in on the same conduit in the building and that conduit floods, freezes, and cracks, both carriers are cut off and connectivity is lost.
Unitas Global has a connectivity solution that uses GIS data, which can see where those fiber routes are physically located in streets and across nine million different buildings. When someone asks for diverse cloud connectivity, we can ensure that entrances into buildings and data centers are fully diverse.
How can businesses map out a disaster recovery plan with their cloud strategy in mind?
- It is necessary first to define the criticality of different services. In a disaster, some components may not be as highly critical to get back up and running as others. Defining an RTO and RPO is also important, which means asking questions like “How long can we be out?”, “How quickly does it need to be up and running?”, and “Can it be up and running with empty data or does it have to be back up with full access to all the historical data?”
- Once you have gathered the necessary information, you can create a plan that starts with the most critical services and then works its way down. Meanwhile, look at the computing and storage infrastructure underneath these services and understand its diversity, as well as the connectivity that connects those components and the diversity there.
- Finally, you want to identify where components physically and logically sit to understand how they communicate with each other. With that information in mind, it is possible to identify specific owners that need to provide restoration of each different component. Typically, this means involvement from network, systems, and software experts, each working on their separate pieces to return each system and service to operation.