MLflow setup and basics in MLOps - Time & Space Complexity

Time Complexity: MLflow setup and basics

O(n)

Understanding Time Complexity

When setting up MLflow and running basic tracking, it's important to understand how the time to log experiments grows as you add more runs or parameters.

We want to know how the system handles more data and if it slows down as usage increases.

Scenario Under Consideration

Analyze the time complexity of the following MLflow tracking code snippet.

import mlflow

mlflow.set_tracking_uri("http://localhost:5000")

with mlflow.start_run():
    mlflow.log_param("param1", 5)
    mlflow.log_metric("accuracy", 0.9)
    mlflow.log_artifact("model.pkl")

This code sets up MLflow tracking and logs parameters, metrics, and artifacts for one run.

Identify Repeating Operations

Look at what repeats when logging multiple runs or parameters.

• Primary operation: Logging parameters, metrics, and artifacts for each run.
• How many times: Once per parameter, metric, or artifact logged; once per run started.

How Execution Grows With Input

As you add more runs and log more data, the total logging operations increase roughly in proportion.

Input Size (n runs)	Approx. Operations
10	About 10 times the logging calls
100	About 100 times the logging calls
1000	About 1000 times the logging calls

Pattern observation: The work grows linearly as you add more runs or log more items.

Final Time Complexity

Time Complexity: O(n)

This means the time to log data grows directly in proportion to how many runs or items you log.

Common Mistake

[X] Wrong: "Logging more runs will only take a little more time, almost constant."

[OK] Correct: Each run and each logged item adds work, so time grows steadily, not fixed.

Interview Connect

Understanding how logging scales helps you design experiments and systems that stay responsive as data grows.

Self-Check

"What if we batch log parameters and metrics instead of logging them one by one? How would the time complexity change?"