Microsoft ‘most accountable’ for service outages last year
Over 13,000 sites were affected by outages last year. Dynatrace’s visual overview of service failures in 2014 gives us a clear picture of which countries and companies they’re coming from.
15.8 failures per hour, 13,116 affected sites and 103.6 years of collective downtime in 2014. How many millions were lost is anyone’s guess. Outages remain a serious issue for all enterprises using third party services. Performance solution provider Dynatrace has used its outage analyser to create a series of charts and graphs on the web failures that occurred throughout last year.
According to Dynatrace’s data, January and February delivered the weakest performance of 2014, each totting up almost double the average amount of hours of downtime last year. The start of 2014 was plagued by several prominent outages, including the Google services interruption, Dropbox’s maintenance glitch and China’s internet outage, all cutting access to millions of users.
In situations of outage, Ops never fail to remind Devs that downtime means losing money, and as Dynatrace aptly puts it “An internet without advertising is an internet without revenue.” So for example when DoubleClick for Publishers went down for five hours on Nov 12th, affecting 3,532 websites in what was “probably the most visible 3rd party service outage ever”, the disruption cost publishers an aggregated $1 million per hour.
Dynatrace’s Wolfgang Gottesheim explained to JAXenter that outages like these are a sad reality:
The outage statistics we have collected over the last year show that outages are not as rare as we would hope, in fact they are a constant fact in our digital world. It shows that we have to expect our external providers to fail, and how problems at a major player like Doubleclick affect a significant portion of the internet. As a developer, I have to be aware of the potential performance impact of third-party services, and need to consider a strategy for addressing performance issues that not only covers my own code, but also includes those external providers.
Outages by region
Dynatrace has grouped last year’s outages geographically:
- US (Pacific Coast): 66,841
- US (Central Mountain): 60,984
- Western Europe: 55,138
- US (Northeast): 50,886
- US (Southeast): 49,117
- East Asia: 31,410
- Canada: 25,495
- US (Great Lakes): 22,526
- Latin America: 22,361
- Australia-Oceania: 15,934
- Eastern Europe and Scandinavia: 15,340
- South Asia: 13,827
- South-East Asia: 13,322
- Middle East: 5,276
- North-Central Asia: 3,269
- Africa: 2,321
Naturally, given the amount of service providers based in the US, the majority of disruptions originated in America. Things become more interesting when comparing the performance of individual companies.
Microsoft still has the hiccups
Microsoft emerges from 2014 as the provider with the most outages, with Azure and Ajax Contend Delivery Network (CDN) responsible for a combined total of 4,602 outages. Google and Yahoo’s analytics and ad services follow in second and third with 4,100 and 3,242 outages respectively.
Although AWS and Amazon’s CDN suffered just over 1,000 outages last year, (putting it in 8th place in Dynatrace’s list), Amazon Data Services outages made up the majority of the single most disruptive outages in terms of duration and impact. Unlike the more dramatic downtime at Facebook, Google and Dropbox, these subtle yet harmful outages have collected little media attention.
Once again, hosting (37.9%) and content delivery (34.3%) services are mainly responsible for the outages. However, Dynatrace’s “Others” category makes up for 27.8% of outages, where advertisting and analytics play a major role.