How Baseline Calculation Works
  • 06 Feb 2024
  • 2 Minutes to read
  • Contributors
  • Dark
  • PDF

How Baseline Calculation Works

  • Dark
  • PDF

Article Summary


Baselines are multiweek rolling averages of monitored data.

Skylight analytics maintains a baseline for every metric of each monitored object. The baseline is a running average of what a metric looks like for a particular hour of the week (for example, 8 PM on Wednesdays). You can visualize this baseline in the object details view.

Baselines appear as a dotted line in line on a per session basis in Inventory.

Tip: You can use baselines to generate Dynamic Thresholds, providing more realistic thresholds that correspond to per-object behavior.

Show or Hide Baselines

Show baselines.gif

To show or hide baselines

  1. Click the Inventory icon in the left sidebar.
  2. Select a session.
  3. Click the Views dropdown list on the upper-right.
    Tip: Access the Views dropdown list from the Overview or Performance tab.
  4. Click the Show/hide baselines icon.
    The icon color indicates whether baselines are shown or hidden:
    • baseline shown - Baseline shown
    • baseline hidden - Baseline hidden

How Baselines Work

Baselines use these attributes:

  • Default Granularity = 3600 seconds / 1 hour
    • All data points within the Granularity window are averaged and stored for each indicator
  • Default Number of Weeks they are kept = 4
    • Baselines are considered reliable after this number of weeks has passed
  • One week (seven days) of baseline data must be accumulated before baseline data can be used in graphs

We do not recommend you set baselines to work at 15m granularity (instead of 1h) as this will impact performance and require substantial reworking of infrastructure on how baseline data is generated/stored.

The default cleaning threshold is set so that 1/60 (1.667%) of dirty data is discarded. Error states automatically trigger data cleaning and cannot be turned off, therefore, having the threshold set so low means you should not use data cleaning as you would render your baselines less and less valuable.

Example Baseline Calculation

Here is an example of calculating Baseline (BL) datapoint for Monday 10 AM for a Metric(M) SMA (Simple Moving Average) over a six-week period:


Mavg is Metric average.
Mavg_Mon_10_w1 is the Metric Average value for Monday 10 am on week 1.

Caution: Changing the Granularity setting deletes all previous baseline data.

By default, baselines are calculated and stored for all indicators collected, but there may be situations in which baselines are not needed, such as:

  • Non-critical Device Groups
  • Unimportant Object Groups
  • Static Indicators (for example, amount of RAM installed in a server)

© 2024 Accedian Networks Inc. All rights reserved. Accedian®, Accedian Networks®,  the Accedian logo™, Skylight™, Skylight Interceptor™ and per-packet intel™, are trademarks or registered trademarks of Accedian Networks Inc. To view a list of Accedian trademarks visit: 

Was this article helpful?

Changing your password will log you out immediately. Use the new password to log back in.
First name must have atleast 2 characters. Numbers and special characters are not allowed.
Last name must have atleast 1 characters. Numbers and special characters are not allowed.
Enter a valid email
Enter a valid password
Your profile has been successfully updated.