Last week marked a return to office for many of us and the restart of infrastructure works for another year.
Outage numbers that were already on a downward trend through December dropped even further across the board over the Christmas and New Year period. Though they have started to rise when compared to the previous two weeks, the base of comparison is very low. Total global outages reported a 15% increase compared to the previous week, which was still 8% lower than the number of outages observed globally in mid-December 2021. This was reflected domestically, where we observed a 10% increase in outages when compared to the previous week, which, similar to global numbers, was 9% lower than outages observed within the U.S. before the holiday period.
While there isn’t much to discuss from the break (something I’m sure most engineers will be relieved about), as we set the foundations for a new year, it’s crucial—but also instructive—to understand what we learned from the past 12 months, and what learnings we can take into 2022.
First, to the figures.
Outages were up year-over-year across the board. The number of cloud service provider outages was almost three times higher in December 2021 than in December 2020. With the number of cloud service provider outages consistently trending higher throughout 2021 and now accounting for around 4% more of all outages observed than they did at the beginning of 2021.
The number of unified communications (UCaaS) outages also appeared to increase month-over-month and ended up doubling over the course of the year.
Sporadically, these outages broke through to directly impact enterprise customers or end users. However, they were often localized in scope, minimizing any global impact, or had no discernible impact at all. In a number of cases, they appeared to be attributed to new infrastructure being brought online to keep pace with the growth in demand for hosted services in a hybrid-work world.
That demand is evident in the frequency of outages recorded. In December 2020, the medium time between cloud provider outages was 1670 minutes. Over the course of 2021, this dropped to 500 minutes, which could be indicative of increased engineering works to support service usage demand and growth.
So, what can we learn from this?
Two clear trends have emerged that will impact the way we monitor and manage network and IT infrastructures this year.
First, 2021 was the year of interdependency.
While we’ve always been cognizant of the impact that interdependencies can have within our own environments, this past year, we saw the real-world effects of interdependencies in third-party owned and operated environments, and the challenges in managing and maintaining experience and performance in this space.
As hybrid ways of working evolve into a more permanent state, so does our reliance on service providers—cloud, unified communications, software-as-a-service, security-as-a-service and more.
Interdependencies between cloud services from a single provider mean that all may become degraded or inaccessible if one goes down. However, it is relatively straightforward to monitor and configure for that.
It’s more challenging when multiple upstream providers are involved, or when a service is assembled from as-a-service components distributed across a number of providers. A software update or invalid route advertisement by one provider can set off a chain reaction of breakages.
We saw this across last year. If we compare the three months leading up to December 2020 with the same three months leading to December 2021, we see an upward trend in the number of outages, and it is possible that increasing interdependencies between services and providers played a large part in that uptick.
The second trend for 2022 is the criticality of digital experience monitoring.
When it comes to unified communications—a video-based meeting, for example—you really only have one chance to get it right, and that’s when the meeting is initiated and as it progresses. It’s a real-time medium: if the performance is poor, then meeting participants have to live with it. There’s no chance to go back and fix it because the time for the call is gone; the need for the meeting has passed.
All of this means that organizations are doing their best to get ahead of the curve and to optimize the performance and experience of digital productivity tools.
As these tools have become a central tenet of business, organizations are more focused than ever on making sure they provide the most optimal digital experience. Over the past year, we saw a 72% increase in collaboration application monitoring, based on the number of tests that enterprises are running against these applications.
It doesn’t mean they are any less focused on other parts of their infrastructures; it’s simply a reflection of the critical focus required in 2022 and beyond on ensuring everyone in the organization has the best digital experience available to them at all times.