On December 7th, AWS experienced a widespread outage. When large public cloud providers have widescale issues, there can be, in non-technical terms, two results. The first is a widescale incident where customers realize they are essentially out of business. The second case may not be as broad, but several services are down. The first one is terrible, but at least there is one element of clarity: Nothing is working. The latter scenario can be far more confusing.
Aite-Novarica’s view is that no one can or should expect perfection from cloud providers, just as they can’t expect it from on-premises data centers. One advantage of using a cloud provider is that everyone can rest assured that restoration is the provider’s top priority. Broadscale AWS outages impact AWS clients as well as the entire Amazon organization, and the holiday season is an important time of year.
A couple of years ago, Microsoft had issues with a cloud telephony product. The problem was visible to Microsoft leadership, as the Microsoft organization uses those telephony services. There is no sugarcoating the impact of the outage for customers in these situations. However, when these issues occur (and they will), the cloud providers must:
Provide clarity. Cloud providers must provide a clear and believable articulation of the root problem after the incident. Most IT professionals know that a clear line of sight can be hard to come by in the fog of war. However, when the fog begins to clear, the cloud provider had better be first with a clear view.
Be timely. Provide clarity quickly. There’s little value in waiting a month to provide insights. To misapply a famous legal dictum of “justice delayed is justice denied,” clarity a month late is just a waste of time. Providers need to understand that organizations are grilling CIOs and CTOs in real-time. Board members, affected business units, and cloud skeptics are demanding answers; IT leadership must be in a position to answer them.
Take responsibility. The provider has to take responsibility unequivocally. One cloud provider in the past explained to a CIO that the company should have implemented its DR execution to remediate the situation it caused. This may be true, but companies don’t expect their cloud providers’ missteps to be a cause for initiating DR processes. That is not taking responsibility.
Explain why the issue will never happen again. Cloud should have a credible plan to explain that they have learned from the incident and made the necessary fixes. Clients will expect this from AWS since it is part of Amazon, and customer obsession is a key leadership principle. Do we believe at this point that Andy Jassy and Adam Selipsky haven’t demanded answers to this exact question? If you are customer obsessed, you owe your customers the same assurances you provide to senior management. Part of customer obsession must be to learn from past mistakes and ensure those missteps don’t repeat.
If any of these points seem harsh, remember that it’s what CIOs and CTOs will have to answer to. Cloud providers are part of the critical infrastructure—a very profitable part. They don’t get immunity.
Cloud technologies will continue to proliferate in the coming years. One of the few headwinds ANG could see is if cloud providers don’t abide by the above expectations. Cloud providers could risk raining on their own parades.
Reach out to [email protected] to discuss this topic further or share your perspectives or experiences during such an event.
Add new comment