By Jordan Feigenbaum, Product Manager, Financial Markets Network, IPC
Across our industry, throughout all industries really, the talk is about the shift to the public cloud. It’s the buzzword of the moment. Amazon, Google and Microsoft are leading a charge to move as much computing power as possible to cloud-driven services, and everyone seems to be following, with good reason.
There are a lot of advantages to using the public cloud, largely around flexibility and simplicity. But along with this shift come a number of inherent risks.
In this two-part series, we’ll explore some of these risks and come back with some ideas around what you can do to protect your business and yourselves.
Accidents will happen.
Your mission critical applications rely on the public cloud. You access the public cloud via the internet. So what happens when your internet connection is down?
There’s resilience in the internet’s amorphous core, but massive infrastructure outages can and do occur. During this recent holiday season, CenturyLink experienced a massive, multi-state outage caused by a faulty network card that lasted for a full two days (1). This outage prevented or hindered the ability to call 911.
Last summer, there were multiple, unrelated fiber cuts that caused major internet outages on both the CenturyLink and Zayo backbones, taking down internet access around the United States for many hours (2). Were you able to get to your public cloud-based computing power?
Those incidents were due to fiber cuts. What happens when the issue stems from a configuration issue? In 2017, there was a 90 minute outage in the US due to an engineer having made a misconfiguration, which led to a route leak issue and created a ripple effect. As a result, providers such as Verizon, Comcast and Spectrum saw their customers unable to access the internet (3). Large route leaks can occur, and the University of Arizona’s Network Research Lab had a project dedicated to detecting and tracking them; they detected between five and 20 large route leaks every year between 2003 and 2009 (4).
While those problems were focused on the United States, North America doesn’t have a monopoly on internet outages. There have been incidents causing internet unavailability in Singapore (5), Australia (6), and in the second part of this series we’ll explore a bit about the DDoS attack on Dyn in 2016, which caused outages across Europe.
Of course, it’s not just the internet that can have outages. Sometimes the public clouds themselves go down. In July 2018, Google experienced an outage to their Global Load Balancers that took down Spotify, Snapchat and other popular services (7). Last year, Amazon saw two distinct power outages in the US East region of AWS in March (8) and June (9).
In Part 2 of this series, I’ll discuss outages and unavailability for other reasons, and come to some conclusions on how you can prepare to keep continuity and connectivity to your mission critical applications over the public cloud.
© 2019 IPC Systems, Inc. All Rights Reserved. The contents of this publication are intended for general information purposes only and should not be construed as legal or regulatory advice.