What good KPI's are there to measure the success of implementing DevOps in an organisation?

202 viewscircle icon2 Comments
Sort by:
Lead Cloud Transformation Architect2 years ago

As Sachin rightly pointed out there is a dependency on the type of application and info you are collecting. In general, the below metrics are some of the key ones tracked based on my experience:

Observability System:
- Application Performance
- Mean Time to Detect (MTTD)
- Mean Time to Diagnose (MTTDi)
- Incident Response Efficiency
- Mean Time to Recover (MTTR)

Application Types (Microservice/Monolith/Infra):
Microservice Adoption:
- Deployment Frequency
- Lead Time
- Change Failure Rate
- End users' Satisfaction

Monolith Modernisation:
- Deployment Frequency
- Lead Time
- Change Failure Rate
- End users' Satisfaction

Infrastructure Optimization:
- Deployment Frequency
- Lead Time
- Change Failure Rate
- Infrastructure Utilization
- Cost Savings

Cross-cutting KPIs:
- Error Rate Reduction
- Compliance and Security
- Release Confidence
- Scalability and Resource Optimization

There are many "It Depends" but this structure ensures a comprehensive coverage of metrics that aligns KPIs with the observability system and the different application types. 

Engineering Manager in Software2 years ago

This will purely depend on the Observability system in place and what information you are logging and tracing in it. Whether your Applications are Micro Service / Monolith / Infra.

Take all the information which you are collected and what is the problem present in your organization. 

Popular Metrics are (Latency, Throughput, MTBF, MTTR and many more)

for example

I have the application ABC and you are getting user requests that sometimes few features don't work and your CIO decides to resolve these issues or similar issues as a top priority and he needs metrics on it ( how many are created and log tickets so that he is aware of it ).

Assume this is Micro Service and you are logging every request and Trace is also recording in it.

As per the User's Complain you have time and logs for the Application and create Metric to find the problem using Log and Trace.

Provide this information to dev teams with details steps to resolve fast and close this issue and keep Metric on HTTP response code (401, 529) as your baseline and continuously create tickets for your dev team.

I hope you got the response you are looking for it. I have taken very base case with very limited information to help you.

Content you might like

Increased efficiency92%

Less demand7%

Need both29%

Storytelling (Competitive advantage, risk mitigation, innovation/agility)47%

Metrics (ROI, TCO, NPV, etc.)24%

Neither

View Results