MLTP

Metrics

• What is it?
• Read World Similarity
• Sample

  metric_name{label="value", ..} measurement
  ...
  histogram_name{label="value", ..} bucket
  

Logs

• Who doesn't know it?
• Collect it to centralised storage, so we can link to our Central Monitoring-UI like Grafana
• Sample Data Model

  log_id: "unique identifier for the log"
  timestamp: "timestamp when the log was generated"
  message: "content of the log"
  severity: "level of severity of the log"
  tags: "key-value pairs for additional information"
  

Traces

• Tracing is a method used to monitor and understand the flow of a request through a distributed system.
• Read World Example of a Trace
• Microservice Example of a Trace - @graph.wealthy
• Sample Data Model

  trace_id: "unique identifier for the trace"
  span_id: "unique identifier for the span"
  parent_span_id: "id of the parent span"
  start_time: "timestamp when the span started"
  end_time: "timestamp when the span ended"
  operation_name: "name of the operation"
  tags: "key-value pairs for additional information"
  logs: "events that occurred during the span"
  

Continuous Profiling

• Continuous Profiling is like a doctor checking your health regularly, not just when you're sick.
• It helps us understand how our system is performing over time, not just when there's a problem.
• Sample Data Model

  profile_id: "unique identifier for the profile"
  start_time: "timestamp when the profiling started"
  end_time: "timestamp when the profiling ended"
  duration: "duration of the profiling"
  cpu_time: "total CPU time used during the profiling"
  memory_usage: "total memory used during the profiling"
  disk_io: "total disk I/O during the profiling"
  network_io: "total network I/O during the profiling"
  

Flamegraph 🔥

• Flamegraph is a visualization tool that presents a graphical representation of the execution of a program.
• It helps us understand the flow of execution and identify performance bottlenecks.
• Google Chrome as example & Demo flame

Golden Signals 🚦

From SRE handbook

• Golden Signals are like the health indicators of a system.
• They help us understand if the system is working well or not.
• There are four main golden signals:
  • Latency: 🕒 How long it takes for the system to respond.
  • Traffic: 🚦 How much data the system is handling.
  • Errors: ❌ How many mistakes the system is making.
  • Saturation: 🔄 How full the system is.