Skip to content

A compression technique motivated by the semantic characteristics of microservices' traces

Notifications You must be signed in to change notification settings

HaseebLUMS/compression-for-distributed-tracing

Repository files navigation

compression-for-distributed-tracing

A compression technique motivated by the semantic characteristics of microservices' traces

Initial results:

Percentage decrease in the number of stored spans = 28%
Percentage decrease in the number of stored fields (key value pairs extracted from logs) = 88%
Percentage decrease in the storage (in MBs) = 70.8%

Steps:

  1. Follow the article for setting up distributed tracing and hosting a microservices site
  2. Visit the hosted site and do some actions that will generate some traces(in future, this step will be automated)
  3. Setup MySQL and create two empty databases: "original" and "compressed" (In future, this step will be automated)
  4. Create an emoty directory results where all the comparison (original db vs compressed db) results will be stored (too lazy to write a line of code for this).
  5. Run run.py 4.1) The traces will get stored in a newly created directory traces 4.2) The two databases will get populated 4.3) The original db will have all the traces in full form 4.4) The compressed db will also have all traces but in the compressed form.

About

A compression technique motivated by the semantic characteristics of microservices' traces

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages