Skip to content
Justin Reynolds edited this page May 30, 2018 · 5 revisions

Vizceral Logo

  1. What is Vizceral?
  2. Why Vizceral?
  3. How do I use Vizceral?
  4. How does Vizceral work?
  5. Other Resources

What is Vizceral?

Vizceral is a visualization component for displaying flows between nodes in a nested set of graphs. Each graph has a set of nodes and connections to be rendered with flow metrics for each connection. Each node can also have it's own set of nodes and connections, allowing the user to 'drill into' each node to visualize another level deep of traffic information.

Netflix currently uses it with three levels: traffic requests into each AWS region, traffic between individual microservices within those regions, and lastly to specifically focus on traffic flowing into and out of a specific microservice.

Why Vizceral?

Vizceral arose out of the need to be able to understand the state of Netflix's complex microservice architecture in any given region, at a glance, when performing a traffic failover. After researching all known available visualization techniques for the amount of data we need to keep our eye on, we decided that there was no solution to fit our needs, so we decided to craft our own. For more about the idea of 'feeling' the state of the system, check out the Monitorama PDX 2016 talk, Intuition Engineering at Netflix, by @jrsquared.

How do I use Vizceral?

For a quick setup, check Getting Started.

For detailed setup, check Configuration.

For more advanced usage, check How to Use.

How does Vizceral work?

Vizceral is effectively just a renderer for a JSON data structure, described in much more detail in How to Use. It takes this JSON structure, runs a layout algorithm on the nodes and connections, and displays the graph on the screen. All other visual features are related to metadata provided within the JSON structure itself.

As far as the metrics themselves are concerned, since we are dealing with metric values that are too large to visualize 1:1 effectively, there is some level of sampling in the number of dots animated. When there are so many things to keep track of and watch, exact numbers stop being as important and relative information is much more actionable.

Other Resources