Sales development, customer satisfaction, research and development, software development... Performance indicators are essential for all core businesses to track progress. But for each profession, the measurement tools vary. In the DevOps world, the main metrics can be grouped under two main headings: Dora and MTTx.
What are the DORA indicators?
DORA (DevOps Research and Assessment) is a research team that has identified 4 performance indicators for software development teams.
Deployment frequency
This is the number of versions released in production over a given period. Since DevOps uses CI/CD for continuous integration and deployment, this performance indicator is generally quite high. And yes, DevOps often release several new versions per hour. This enables problems to be detected quickly, and corrected immediately.
Delivery time
As a corollary to the previous point, this DORA indicator corresponds to the period between a code modification and its release into production. But beware, delivery time does not always follow deployment frequency.
Indeed, code is often deployed in different branches, with sub-branches for feature additions, bug fixes… and a main branch corresponding to production release.
As a result, certain modifications may remain pending in sub-branches without being put into production. In other words, without being presented to users. This latency is measured by the delivery time.
Modification failure rate
The aim is to measure the percentage of deployments that fail to go live. Measuring these failures is essential, as they affect both end-users and development teams.
For the latter, each failure may involve a cancellation of the deployment, a correction, a modification, and so on. So much time spent solving a problem rather than developing new features.
It is therefore essential to reduce this failure rate, but not to bring it to zero, at the risk of being over-cautious.
Service restoration time
The aim of this Dora indicator is to measure the time needed to recover from a production failure. In other words, until end-users are no longer affected by the bug.
Monitoring this metric is all the more important as it directly affects customer satisfaction, and therefore the commercial success of the software or application.
For DevOps teams, the challenge is to minimize this service restoration time, thanks to a robust infrastructure that remains functional, despite bugs during development.
What are MTTx metrics?
Alongside the DORA indicators, there are MTTx metrics that focus on operations.
MTTD
This MTTx performance indicator focuses on problem detection. How long does it take before DevOps teams detect a bug or error?
Ideally, the MTTD should be as short as possible, to enable developers to correct the bug quickly, and above all, to avoid the accumulation of problems.
To achieve this, it’s essential to set up an effective alert system, but also to launch deployments regularly.
MTTA
Mean time to action is the time taken between the detection of an anomaly and the action taken to correct it. This MTTx indicator primarily measures the responsiveness of DevOps engineers in resolving a problem.
Here again, the shorter the MTTA, the more efficient the teams, and the greater the end-user’s satisfaction.
MTTC
The letter C stands for clue. In this case, it’s a question of evaluating the time needed between detection of the problem and identification of its cause. If this MTTx indicator is too high, it means that the software is too complex.
Teams have too many things to manage. They lack the visibility to effectively understand the entire development environment. This makes solving the problem much more time-consuming.
In this case, it’s vital to simplify the process as much as possible and automate certain tasks to free up engineers.
MTTR
MTTR stands for Mean Time To resolution. In other words, the MTTx KPI refers to the DORA indicator for recovery time.
This latter indicator is totally dependent on the two preceding ones. If the time to act and/or identify the cause is too long, teams will spend much longer resolving the problem.
How do I use Dora and MTTx metrics?
To measure Dora and MTTx indicators, you first need effective observability and monitoring tools. But above all, the aim is to optimize the performance of DevOps teams.
To achieve this, a number of measures need to be put in place, such as automation, definition of common objectives, continuous deployment and testing.
To find out which best practices to implement right from the start, it’s best to get trained. With DataScientest, you’ll develop a theoretical and practical understanding of the best development methods. Find out more about our program.
