Skip to content

Latest commit

 

History

History
104 lines (47 loc) · 3.58 KB

README.md

File metadata and controls

104 lines (47 loc) · 3.58 KB
Raw

#PerfSONAR Control Plane (pscp)

Our control plane is used to start a test between two perfSONAR hosts, monitor the throughput performance and locate the source of network issues when problem happens.

1.Preparation Work: offline Measurement Archive (MA) data collecting

1.1 ma_filter.py

This script can be used to query and save the list of all MA hosts, and then uses ping to move out the hosts that can't be reached. After running this script, two files are generated: ma_list and filtered_ma_list.

In this script, we call a module in query_diy.py, which will return a list of MA hosts from perfSONAR lookup service directory.

1.2 traceroute_collector.py

This script is used to build a traceroute dataset by retrieving the traceroute tests records from MAs of each host listed in "filtered_ma_list" using perfSONAR client REST interface.

To start collecting traceroute data from MA hosts, use command:

    python traceroute_collector.py

2.Online Control Plane

control_plane.py is the code for online measurement test controls. We implement the meaausrement and troubleshooting methods between two specific perfSONAR nodes.

To start the control plane, use command:

 python control_plane.py [source_host_name] [destination_host_name]

In control_plane.py, we deploy the following three modules: bwctl_tools.py, find_ps_node.py and issue_locator.py

2.1 bwctl_tool.py

This module is used to start a throughput test (use bwctl tool) or a traceroute test (use bwtraceroute tool) and return the result.

2.2 find_ps_node.py

After obtaining the traceroute of a problematic path, this module is used to find the nearest perfSONAR node for each target router on the path by looking up the graph we built.

2.2.1 build_graph.py

The objective is to build a traceroute graph for find_ps_node module when we detect the abnormalities.

2.2.2 graph_example.py

This is a small example of replacing the IP address of target routers on a specific path with perfSONAR nodes' IP.

2.3 issue_locator.py

An example of problematic path:

  A-----C--X--D-----E-----F-----I-----H-----G-----B

After replacing the routers with their nearest perfSONAR nodes, this module is used to start throughput tests from both end nodes(A, B) to each other nodes to find the long-clean path (B-D, RTT>20ms), and locate the source (C-D) of network issue .

3.Other Tool

3.1 traceroute_recorder.py

We collect all the traceroute test URLs from MAs and write the records into a file. Later, this file will be used to update our traceroute detail dataset as well as the traceroute graph.

3.2 ma_filter_example.py

It is used for collecting active nodes from file ma_list[data].

3.3 graphviz_example.py

It is used for visualizing the traceroute graph. It needs graphviz Python library.

3.4 threshold_determine.py

It is used for determining the threshold to identify if the path has a problem by looking into the historical bandwideth test results between these two ndoes. If there is no test result before, we will use the default threshold that is 1Gbps.

4.Dataset (json format)

4.1 ma_list[date]

It is a dataset contains all the MAs' hostname.

4.2 filtered_ma_list[date]

It is a dataset after we filter the "bad" MAs in ma_list.

4.3 ma_record_[timestamp]

It is a dataset contains all the traceroute tests uris.

4.4 dataset

It is a dataset contains the details of all the traceroute tests.

4.5 dataset95

It is a small dataset we queried from 95 chosen MA hosts in central and eastern regions of US.