Stephen Elliot and the IDC team recently released a report (registration required) sponsored by AppDynamics that examines the true cost of downtime and infrastructure failure, as well as DevOps adoption. The headline numbers are pretty eye-opening:
- For the Fortune 1000, the average total cost of unplanned application downtime per year is $1.25 billion to $2.5 billion.
- The average hourly cost of an infrastructure failure is $100,000 per hour.
- The average cost of a critical application failure per hour is $500,000 to $1 million.
Let those numbers just sink in a moment. Between $1 to $2.5 billion lost to application downtime a year! That is a big number, but if you look at the follow on numbers you can see how you can get there. Infrastructure failure has an average hourly cost of $100k per hour, critical application failure has an average hourly cost of $500k to $1m per hour. It doesn’t take a lot of hours for those numbers to add up. You can read in the report the actual math IDC applied to reach these conclusions based on the responses to a survey they undertook.
There was also some good insight into DevOps and DevOps tools adoption. Remember this was aimed at the Fortune 1000 market segment, so large enterprises. About 43% of respondents were using DevOps currently, with another 40% planning on adopting DevOps soon. That 83% number is in line with another recent survey by CA that showed an 88% number.
But all is not as rosy as that picture paints. When looking at the biggest impediments to DevOps, the usual suspects rear their ugly heads:
- Cultural inhibitors – 56.7%
- Fragmented processes – 43.3%
- Lack of executive support – 26.7%
All three of these topics have been the fodder for stories here on DevOps.com. There are specific strategies to overcome each of these, but the fact is they are still major obstacles to successful DevOps adoption despite overwhelming desire to do so.
There was another interesting metric in the study that bears noting as well. It is about companies who try to use their existing tool sets in a DevOps setting. There is something like an 80% failure rate for companies who try to adapt their existing tools for DevOps practices. That screams out something that many a man already knows, “you need the right tool to do the job right.” The moral of that story is if you are going DevOps you damn better be buying some DevOps specific tools to get the job done right.
Again in line with other surveys I have seen the biggest initiatives driving DevOps adoption are:
- Automation 60%
- Continuous delivery 50%
- Continuous integration 43.3%
- Automated testing 43.3%
- Application monitoring/management 43.3%
DevOps was being initiated by slightly more development teams that IT Ops teams by about a 40% to 33% margin respectively.
You can download the full report from the link at the top of the story to get the rest of the data. Ultimately like similar surveys there are some assumptions that you have to swallow in order to give credence to the findings. For me the biggest grains of salt to consider are:
- This survey was based on a relatively small amount of respondents. On average like 50 per question and some of those not from organizations in the scope of the survey. With that small a group trends could be skewed by a handful of outlier responses.
- While IDC is a very reputable company, the fact is this was a sponsored survey by a DevOps tool vendor. AppDynamics likewise is a reputable company (and in full disclosure a sponsor here at DevOps.com), nevertheless it is something you should take into consideration.
- Perhaps the biggest grain is: Do you buy their math in calculating downtime? According the report’s authors here is how they arrived at their numbers:
To examine application downtime impact per year across the Fortune 1000, we conducted the
- 1,000 companies multiplied by an average of 250 critical applications per company equals 250,000 total applications. 250,000 multiplied by an average of $500 to $1 million per hour of downtime cost equals $1.25 billion to 2.5 billion of unplanned application downtime costs per year.
That sounds like back of napkin stuff to me. But really at the end of the day whether the number is $1b, $2b or more, suffice to say it is a big number.
Surveys like these continue to contribute to a broader understanding of the DevOps market. It also provides cover to IT executives seeking support to kick off DevOps in their own organizations. Hopefully we are approaching a critical mass of data that will make the job of “selling” DevOps up the org chart an easier sell.
Here’s a great upcoming webinar that will include thought leadership from Stephen Elliot, VP of analyst firm IDC on “DevOps and the Cost of Downtime” based on research of Fortune 1000 companies. Register for it now, here.