Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Description of Faban data #52

Open
VincenzoFerme opened this issue May 10, 2016 · 1 comment
Open

Description of Faban data #52

VincenzoFerme opened this issue May 10, 2016 · 1 comment
Assignees
Milestone

Comments

@VincenzoFerme
Copy link
Member

VincenzoFerme commented May 10, 2016

Faban collects data about performance from the harness/master and each of the utilised agents. The number of agents utilised is arbitrary (from 0 to N), as well as the use of the harness/master to generate the load is optional. That said, at least one among harness/master and agents must be utilised.

For each of the previous harness/master or agent, Faban produces a set of file. In the following I describe the files that are of our interest. Whenever I refer to fields and names that seems not defined, you can find their definition in http://faban.org/1.3/docs/guide/harness/results.html. The files are located in the /app/faban/FABAN_RUN_ID folder for each trial. An overview of the different information that are present can be accessed here: http://faban.org/1.3/docs/guide/harness/results.html.

I attach a zip file containing one of the folder (name: web101.2Q) I mentioned before, with files generated using the master and one agent.

The RunInfo (files: resultinfo, summary.xml*)

Files

  • resultinfo is just a single file containing the Status field
  • For each harness/master and agent you find a summary.xml.* file
  • Then you have a summary.xml file that aggregates the information of all the summary.xml.* files

Structure of the file

  • resultinfo is plain text, tab separated file
  • summary.xml.* are xml files with a clear structure that is self-explicative. As all the xml files in Faban, they refer to an xslt file (useful to get hints on how to parse these files). I attach them in the xslt zip file. The information relevant for the RunInfo are defined in the benchSummary entity, some of them in attributes, some other as fields.

What to get from these files

  • The RunInfo information as per in http://faban.org/1.3/docs/guide/harness/results.html for each of the harness/master and agent (which name is identified by the host attribute of the benchSummary entity), and the general and aggregated RunInfo where the host should be set to aggregate.

The Summary Report (files: summary.xml*)

Files

  • For each harness/master and agent you find a summary.xml.* file
  • Then you have a summary.xml file that aggregates the information of all the summary.xml.* files

Structure of the file

  • summary.xml.* as in the previous section. The drivers information are stored in the driverSummary entity.
  • Moreover for each driver (the number can be arbitrary) you find the following entities: mix, responseTimes, delayTimes, [customStats]. customStats is optional and we can arbitrary have a number from 0 to N of these entities, each one uniquely identified by the name. The structure of these entities is straightforward, just refer to the name of the fields and attributes.
    • mix: you can have an arbitrary number of operation entities
    • responseTimes: you can have an arbitrary number of operation entities and fields. The structure of the content, especially for what concern the percentiles is dynamic
  • To get information about the "Little's Law Verification" you need to refer to the xlst file because it's computed there. It should be the rtXtps field for each driver.

What to get from these files

  • The Summary Report information as per in http://faban.org/1.3/docs/guide/harness/results.html for each of the harness/master and agent (which name is identified by the host attribute of the benchSummary entity), and the general and aggregated RunInfo where the host should be set to aggregate.

The Detail Report (files: detail.xan.*)

Files

  • For each harness/master and agent you find a detail.xan.* file
  • Then you have a detail.xan file that aggregates the information of all the detail.xan.* files

Structure of the file

  • Short description of the files, and the use to the stored data in Faban: https://perfwork.wordpress.com/2012/08/27/upcoming-new-features-in-faban/
  • They are text file where each set of data it's identified by a Section, the type of visualisation by a Display field and the data are stored in a table. The number or rows is arbitrary, as well as the number of columns because it depends on the number of operations. The unit of measure of the timing columns its also arbitrary and its identified in the name of the column itself

What to get from these files

  • The Detail Report information as per in http://faban.org/1.3/docs/guide/harness/results.html for each of the harness/master and agent (which name is identified by the host field in the file), and the general and aggregated Detail Report where the host should be set to aggregate.

Related issue: benchflow/collectors#36

@Cerfoglg feel free to ask questions so that I can improve the description.

web101.2Q.zip
xslt.zip

@Cerfoglg
Copy link
Contributor

@VincenzoFerme Faban transformers and analysers are now implemented

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Development

No branches or pull requests

2 participants