Feedback loops are critical for DevOps success. Knowledge of how changes affect users -- and how they're received -- helps a DevOps team shape its approach throughout the application lifecycle.
To work, feedback loops require data. This data comes primarily from end users' suggestions and help desk tickets, but a comprehensive monitoring tool is just as important to understand a DevOps environment's health.
It's critical to monitor service availability to provide better support and turnaround times when issues arise. Basic monitoring normally includes a variety of alerts for events such as a server going offline or a database becoming unavailable. These alerts' relevancy depends on an IT environment's topology, and high availability and redundancy configuration, but it's useful to know something is off when it should be on.
Skim the waters with end-to-end monitoring
Rather than a system that evaluates each component individually, DevOps teams can use an end-to-end monitoring tool. These tools watch over the UX and return data such as wait time between request and resolution.
One particularly useful feature is the ability to see what stage of user request suffered a significant delay: Rather than knowing it took 10 seconds for the user to see a response, an IT admin can see that IIS Server 27 took eight seconds to get a response from SQL Server 2. This detail enables operations staff to communicate clearly any end-user experience issues they see, and funnel those issues to the most appropriate team.
Avoid -- at all costs -- the blame game in issue troubleshooting. This is when help desk knows about a widespread problem, and the systems administrator blames the network team, who blames the firewall and security team, who denies responsibility and reroutes the problem to the database admins. This shuffle is a massive waste of time for all parties -- especially the user.
IT operations should have visibility into a production environment not to assign blame, but to identify potential weak points and underlying bottlenecks or inefficiencies. This visibility, for example, enables IT operations admins to notice that a SQL cluster is the source of delays across multiple applications, or that a specific SQL server in a cluster continually displays errors or performance degradation.
IT operations is the driving factor that keeps communication lines open: The ops specialists in a DevOps organization must be able to oversee the environment without having to deep dive into each aspect of application development and deployment, or wade through logs for a hint of a problem. An end-to-end monitoring tool provides the necessary information to get teams to talk and prevent groups from hiding their own major issues.
An end-to-end monitoring tool or tool suite is useful to identify, at a glance, IT issues and their causes. IT operations can apply these monitoring tools for both service outages and degradations to increase efficiency.
Dig Deeper on Real-Time Performance Monitoring and Management
Related Q&A from Adam Fowler
Outlook crashes and rendering issues might require an upgrade -- or a downgrade -- for administrators who want to avoid a Band-Aid approach and want ... Continue Reading
I have nothing but an IT help desk email inbox, and deployment management is organized chaos. It works well enough to keep the business running, but ... Continue Reading
What are some ways to level up from an IT operations or administration role into an IT engineering one? Will my experience in IT operations ... Continue Reading