You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 4 Next »

Project Title:

IEPM-BW Connectivity Visualization

Author: Les Cottrell 

Personnel:

Asif Khan will put together the original version, then an NIIT student, under Asif's guidance will study the documentation and implementation and provide feedback, and then respond to user requests and turn into a finished product.

Project Aim:

The aim of this project is to provide a broad top level visual overview of the ping connectivity of the IEPM-BW remote hosts as seen from an IEPM-BW monitoring host.

Motivation:

IEPM-BW monitoring hosts currently make ping measurements to about 40 remote hosts. These measurements are made at ten minute intervals, and include a set of ten default length pings to each of the remote hosts. The number of remote hosts is expected to increase to about 100 or so. We need a way for a user to quickly ascertain the reachability of the remote hosts over some time period. A remote host is deemed unreachable if all (typically 10) pings in a set fail. We can also extend the technique to show not just reachability but also losses or Round Trip Times (RTTs).

Project Description:

The IEPM-BW ping data will be read from the IEPM-BW Measurement Archive (MA) by a web services interface. Asif has put together a first cut that gets the data and does a simple HTML table display, from inside SLAC it can be viewed via http://net-desk1/traceanal/ping_analysis.pl?hours=12. Asif will supply information on how to access the first cut code. It should be possible to specify the time interval for which the data is required, as well as the remote host. It should also be possible to specify all available remote hosts. The data returned will contain at least the name of the remote host, the time stamp of the measurement, and the number of pings sent and received. It should also be possible for the data to contain RTT.

The simplest display will have the name of the remote host in the first and last columns. Each of the other columns will be for different time stamps (i.e. 6 columns/hour if the measurements are taken at 10 minute intervals). Each row (apart from the header and footer which contain the time stamps) is for a remote host.

For the losses, the contents of a table cell will be typically be green if no pings are lost, red if all pings are lost (unreachable), and a gradation of colors between red and green for intermediate losses.

For RTTs (selected by means of a check box) the colors will indicate the RTTs. Red for > 550ms, and a gradation of colors (e.g. light red, magenta, pink, orange, yellow, yellow-green, green, cyan) to blue for < 10ms RTT (we will probably need to experiment with this).

The table should be as compact as possible in order to provide visualization of as many nodes and timestamps as possible. One could imagine a cell being one pixel wide so that one could display several days worth of measurements in a 1280 pixel wide screen. In such a case labeling the timestamp for every measurement would be a challenge so maybe one would only label the time every 8 hours or so.

Mouseover a hostname will provide details on the host (Connie and or Jerrod can make suggestions for what should be displayed), clicking on the host would provide access to the IEPM-BW time series measurements, e.g. to http://www.slac.stanford.edu/comp/net/iepm-bw.slac.stanford.edu/slac_wan_bw_tests.html#iepm-bw.bnl.org or to http://www.slac.stanford.edu/comp/net/iepm-bw.slac.stanford.edu/toolrpts/html/iepm-bw.bnl.org.html, this will need to be parameterized for the monitoring host.

Mouseover a cell could provide the actual measurement time, or the actual ping measurement if available, or just the textual loss and RTTs (min/avg/max), or whatever is available. Clicking on the cell will take you to ping time series plots for the selected host.
We will probably also want a history table such that there is one column per day. Red in this case would mean the host was unreachable all day, green that it was reachable all day, intermediate colors will be for intermediate unreachable states (i.e. not for losses). One might have as an alternative (or as a check box option) the colors to mean average loss over the day, red in this case would be 100% loss for the day (= unreachable all day), green for no loss etc. I am not sure what to display for mouseover. Click would take one to the relevant history plot (e.g. http://www.slac.stanford.edu/comp/net/iepm-bw.slac.stanford.edu/html/histplts/iepm-bw.bnl.org.html).

The help should give a legend of the meanings for the colors. There might also be a link to relevant traceroutes.

 Suggestions for improvements

  • It needs a help entry that provides help (e.g. what the colors  mean, how to use various features) when one does a mouseover.
  • It needs a checkbox to select to display RTT rather than losses, or an alternate might be to try two colors per entry one for losses the other for RTT.
  • It needs the node names displayed in the right hand column.
  • It needs mouseover and clicking on hostname.
  • Need to experiment with more compact format.

Requirements:

The student will need to be or become proficient in Unix/Linux, perl, CGI scripts and using web services. Some knowleedge of databases is desirable. The code will need to be production quality. Guidelines on how to write perl will be provided. The student will need to apply for and get a Unix account at SLAC and will be provided access to the relevant computers, files and databases.

  • No labels