<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=us-ascii">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.7226.0">
<TITLE>Scalable Monitoring - RFC</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/rtf format -->
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">Hi All,</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">Following the path forward</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">requirements</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">review by Matt</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">L. in the last OFA Dev Summit</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">I have</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> </SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">started thinking what would make a monitoring system scale to tens of thousand of node.</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">This RFC provides both</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">what I</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">propose as</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">the requirements list as well as a draft implementation prop</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">o</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">sal - just to start the discussion.</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">I</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">apologize</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> for the long mail but I think this issue deserves a careful design (been there done that</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">…</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">)</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><B><U></U></B></SPAN><SPAN LANG="en-us"><B><U><FONT COLOR="#0000FF" FACE="Calibri">Scalable fabric monitoring requirements:</FONT></U></B></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">* scale up to 48k nodes</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> </SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">* 16 ports which gets to about 1,000,000 ports.</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> (</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">16 ports per device is</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">average</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> for 32 ports for switch and</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">1</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> fo</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">r</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> HCA</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">)</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">* provide alerts for ports crossing some rate of change</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">* support profiling of data flow through the fabric</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">* be able to handle changes in topology due to MTBF.</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><B><U></U></B></SPAN><SPAN LANG="en-us"><B><U><FONT COLOR="#0000FF" FACE="Calibri">Basic design</FONT></U></B></SPAN><SPAN LANG="en-us"><B><U></U></B></SPAN><SPAN LANG="en-us"><B><U> <FONT COLOR="#0000FF" FACE="Calibri">considerations</FONT></U></B></SPAN><SPAN LANG="en-us"><B><U></U></B></SPAN><SPAN LANG="en-us"><B><U><FONT COLOR="#0000FF" FACE="Calibri">:</FONT></U></B></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">* a distributed</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">data</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">collection scheme is required. No single manager can go through all the ports in reasonable time.</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">* the system should push as much work to the agents. Examples for that are rate calculation and counter resets.</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> </SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">* features like data compression or aggregation are important for reducing the data</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">reported</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">up</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">-the-tree to</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">the console or data storage.</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">To support that the agents should be able to:</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">1. Report error counters increase only if crosse</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">s</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> a</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">programmable</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> threshold</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri"> 2. Aggregate data and packet counters. Provide upstream data only when change is larger then a given threshold</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> </SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">3. Create and provides histogram of rate of change for each counter</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> (merging data for all ports it collects counters for).</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">* To support data profiling we will need to store xmit recv data for each port which boils down to huge amount of data.</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> </SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Instead of rolling that data up</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">-</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">the</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">-</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">tree each agent should be able to write its own file and merge offline.</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">* handling topology changes is a challenge for a distributed set of agents. Two problem arise from topology dynamics:</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">1.</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">An agent</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">responsible</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> for query of some device might loose connection to it.</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">So</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">dynamic responsibility assignment is</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> </SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri"> </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">required to support dynamic topology. </FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">2.</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Agents are arranged in reporting hierarchy which by itself can be disconnected.</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><B><U></U></B></SPAN><SPAN LANG="en-us"><B><U><FONT COLOR="#0000FF" FACE="Calibri">Implementation proposal:</FONT></U></B></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">This section describes a specific agent behavior.</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> </SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">The actual communication i</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">mplementation can be based on IB verbs or over plain sockets.</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><U></U></SPAN><SPAN LANG="en-us"><U><FONT COLOR="#0000FF" FACE="Calibri">Definitions:</FONT></U></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><B><I></I></B></SPAN><SPAN LANG="en-us"><B><I><FONT COLOR="#0000FF" FACE="Calibri">Responsibility Subnet</FONT></I></B></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">: the set of nodes the agent is monitoring</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><B><I></I></B></SPAN><SPAN LANG="en-us"><B><I><FONT COLOR="#0000FF" FACE="Calibri">Responsibility Subnet Perimeter</FONT></I></B></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">: the nodes connected to the</FONT></SPAN><SPAN LANG="en-us"><B><I></I></B></SPAN><SPAN LANG="en-us"><B><I> <FONT COLOR="#0000FF" FACE="Calibri">responsibility subnet</FONT></I></B></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> which are in the responsibility of other agents</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><B><I></I></B></SPAN><SPAN LANG="en-us"><B><I><FONT COLOR="#0000FF" FACE="Calibri">Peer Agents</FONT></I></B></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">: the agents that are responsible for the nodes on the perimeter of the current agent</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><B><I></I></B></SPAN><SPAN LANG="en-us"><B><I><FONT COLOR="#0000FF" FACE="Calibri">Master Agent</FONT></I></B></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">: the agent the current agent should report to. This is configured on the command line.</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><U></U></SPAN><SPAN LANG="en-us"><U><FONT COLOR="#0000FF" FACE="Calibri">Agents data structure:</FONT></U></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">1. </FONT></SPAN><SPAN LANG="en-us"><U></U></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">List of node GUIDs it is handling</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">2. </FONT></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Last read</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">values for all</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">counters for each port it is handling</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">3. </FONT></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Change rate histogram for each port it is handling</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">4. </FONT></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">List of other agents that are responsible for nodes on the perimeter of its responsibility networ</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">k</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<UL DIR=LTR><UL DIR=LTR>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
</UL></UL>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><U></U></SPAN><SPAN LANG="en-us"><U><FONT COLOR="#0000FF" FACE="Calibri">Agents Communication:</FONT></U></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">A broadcast group is used to broadcast messages regarding agents node ownership.</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">Point to point communication is also used as much as possible</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> as described below.</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">The messages involved are</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> (I list data of the message in <>)</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">:</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">1. </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Query: Who monitors <NodeGUID>?<BR>
Response: <Agent IP>, <port num>, <hops> (the distance the agent is from the monitored node), <num</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">nodes</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">owned></FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">2. </FONT></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Trap: <time>, <NodeGUID>, <port num>,</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri"><counter>,</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri"><value>,</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri"><threshold</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> crossed</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">></FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">3. </FONT></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Data: <time>, set of port counters to report.</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">The message is v</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">ariable length.</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">4. </FONT></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Get Histograms: an agent is requested to provide back histograms for all its managed ports</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">5. </FONT></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Clear all</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">: instruct an agent to clear all counters and start fresh</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><U></U></SPAN><SPAN LANG="en-us"><U><FONT COLOR="#0000FF" FACE="Calibri">Agent behavior is:</FONT></U></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">1. </FONT></SPAN><SPAN LANG="en-us"><U></U></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Register to the monitor agents broadcast group – I.e. have a UDP socket listening on it</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">2. </FONT></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Discover the fabric connected to it (in BFS manner while tracking the</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">number of</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">hops):</FONT></SPAN></P>
<UL DIR=LTR><UL DIR=LTR>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">For each node discovered :</FONT></SPAN></P>
<UL DIR=LTR><UL DIR=LTR><UL DIR=LTR>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">If already knows its responsible master – skip</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">Query for responsible master</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> (broadcast</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">“</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">Who Owns?</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">”</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">)</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> If gets response </FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> If same or lower hops then the current hop count add to other nodes responsibility list</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> Else </FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> it is this agent responsibility – notify every other agent by broadcasting this info (response for</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">“</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">who</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">owns?</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">”</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">)</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> add to BFS next step list</FONT></SPAN></P>
</UL></UL></UL></UL></UL>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">3. </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Foreach node in its responsibility list</FONT></SPAN></P>
<UL DIR=LTR><UL DIR=LTR>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">read all counters every</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">“</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">epoch</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">”</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">(means a constant scan</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">time</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">)</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">calculate rate of change for every counter</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">report traps if the counter is error counter or data set at the end of the epoch changed data which crosses the change threshold</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">clear counter overflows</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
</UL></UL>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">4. </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Once in some timeout </FONT></SPAN></P>
<UL DIR=LTR><UL DIR=LTR>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> For each peer in the pe</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">e</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">rs list</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> If peer does not answer run step 2 again</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
</UL></UL>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">5. </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">Respond to incoming messages:</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">a. </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">“</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">Who owns?</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">”</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">-</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">use</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">point t</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">o point</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">b. </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">“</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">Trap</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">”</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">–</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> propagate up the tree</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">c. </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">“</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">Data</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">”</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">–</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> propagate up the tree</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">d. </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">“</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">Histogram</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">”</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">–</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> merge with other child histograms.</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">e. </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">“</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">Get Histogram</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">”</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">–</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> send downwards and after merging sent</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">“</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">Histogram</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">”</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">up</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">.</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">f. </FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">“</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">Clear all</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri">”</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"> <FONT COLOR="#0000FF" FACE="Calibri">–</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><FONT COLOR="#0000FF" FACE="Calibri"> clear self and send downwards</FONT></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"><B></B></SPAN><SPAN LANG="en-us"><B></B></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><B><I></I></B></SPAN><SPAN LANG="en-us"><B><I></I></B></SPAN><SPAN LANG="en-us"><B><I></I></B></SPAN><SPAN LANG="en-us"><B><I></I></B></SPAN><B><I><SPAN LANG="en-us"></SPAN></I></B><A NAME=""><B><I><SPAN LANG="en-us"><FONT COLOR="#0000FF" SIZE=6 FACE="Monotype Corsiva">Eitan Zahavi</FONT></SPAN></I></B></A><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" SIZE=2 FACE="Tahoma">Senior Engineering Director, Software Architect</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" SIZE=2 FACE="Tahoma">Mellanox Technologies LTD</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" SIZE=2 FACE="Tahoma">Tel:+972-4-9097208<BR>
Fax:+972-4-9593245</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT COLOR="#0000FF" SIZE=2 FACE="Tahoma">P.O. Box 586 Yokneam 20692 ISRAEL</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN><SPAN LANG="en-us"></SPAN></P>
</BODY>
</HTML>