Advanced use of IT observability speeds up problem resolution

Over two-thirds of organisations that have mature IT observability say it’s helping them reduce application downtime significantly, compared with less experienced organisations.

The Enterprise Strategy Group poll of 1,850 IT operations staff found that people who consider themselves leaders in observability can resolve IT issues 2.8 times faster than beginners.

The poll, commissioned by Splunk for its State of observability report, found that nearly half (47%) of survey respondents have used observability tools for two years or longer, up from 36% a year ago.

Speed gives leading organisations an edge in software development velocity. Some 76% of the IT operations staff who said their organisation was a leader in observability deploy the majority of their application code on demand, in contrast to 30% of beginners.

“Building a leading observability practice means being obsessed with delivering incredible digital experiences to your customers, and embedding that mindset into every decision,” said Patrick Lin, senior vice-president and general manager of observability at Splunk. “Our report shows this mindset pays off. Leaders not only achieve greater success in mitigating downtime, they also see greater developer innovation and speed.”

The survey reported that engineering teams at leading organisations achieve a 22% higher change success rate for production application code.

According to Splunk, the majority of leaders say these changes are successful 90% of the time or more, which gives developers more time to experiment and innovate. Compared with their peers at those organisations starting out with observability, the survey found that developers in leading organisations spend about 38% more time on innovation versus routine tasks such as maintenance, alert handling and configuration.

Over half (57%) of the IT operations staff polled say they are finding the root cause of problems significantly faster than a year ago. Leading organisations estimate that upwards of 80% of alerts are legitimate. However, in organisations that are just starting out with observability, over half (54%) of alerts are tied to a real issue. According to Splunk, this leaves a lot of room for conjecture, which, after a while, means engineers may start to ignore alerts altogether.

Leading organisations are 2.3x more likely than beginning organisations to measure their mean time to resolve (MTTR) in minutes or just a few hours, while beginning organisations are 2.4x more likely to measure their MTTR in days, weeks or even months.

According to Splunk, the difference between hours and days is a particularly wide chasm, given that it’s estimated to cost organisations $540,000 per hour.

Nearly all survey respondents (97%) use artificial intelligence (AI) and/or machine learning (ML)-powered systems to enhance their observability operations – a significant jump from 66% of respondents surveyed last year.

Splunk said IT teams can use AI and ML to analyse and process large volumes of data to detect anomalies, identify root causes, recommend actions and automate tasks.