[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference netcad::hub_mgnt

Title:DEChub/HUBwatch/PROBEwatch CONFERENCE
Notice:Firmware -2, Doc -3, Power -4, HW kits -5, firm load -6&7
Moderator:NETCAD::COLELLADT
Created:Wed Nov 13 1991
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:4455
Total number of notes:16761

2425.0. "Repeater900TM collision readings??" by SNOC02::BOYLEBRENDAN (Brendan Boyle) Sun Jun 25 1995 22:05

    In a newly installed network, Hubwatch for Windows V3.1 is registering
    a significant collision reading for the majority of ports. The network
    is functioning fine under the low utilisation levels, however, I found
    these collision readings quite hard to explain.
    
    The collision dials for the active ports at the repeater summary screen are
    showing red with levels anywhere up to 15%. 
    However, at the port screen the collisions/sec dial is showing very 
    occasional collisions. 
    The utilisation on each port is barely registering, with total load on the
    thinwire port showing < 5%(at both repeater summary and port summary
    screen).
    
    Polling is on at 10 second intervals.
    
    The configuration is somewhat unusual but appears to be legal:  
    
    	Two DEChub900 backplanes populated with 9 x DECrepeater900TMs
    running firmware 1.0. Hub1 is connected to Hub2 via one of the UTP
    repeater ports. All connections made to default thinwire in hub. Some
    DECserver100s are connected via transceiver.
    
    What could be causing this collision reading at such low levels of
    utilisation?? The colision levels do not appear to be consistent in magnitude
    but are consistence in presence. Could this "cascading" of repeaters be
    causing a problem??
    
    BTW. An iteration of the find duplicate address function lists all
    active ports as duplicates, I believe due to the cascade.
    
    Thanks for any assistance/suggestions,
    
    Brendan Boyle
    Network Services, Sydney.
T.RTitleUserPersonal
Name		DateLines
2425.1Try Several RefreshesNETCAD::DRAGONMon Jun 26 1995 15:1125
    
    Brendan,
    	
        What HUBwatch is displaying is the percentage of collisions vs
        total transmit attempts. For example, if you perform a [Refresh] on 
        the Port Summary View HUBwatch gets a sample of total frames, 
        total collisions, total frames in error (including runts). HUBwatch
        then gets some other Port Summary details and collects a second
        sample. The delta time between samples in this case will usually 
        be pretty small. If you have low utilization, the delta time is
        small, and you happen to perform the [Refresh] (or a poll is done)
        when collisions are present, you may end up with a Collision %
        which is scary because its high and not easily explained because 
        the utilization is so low (as in your case).
    
        I tried this on our production network and get very similiar 
        results. What I would be concerned with is if on the Port Summary
        View for a port you consistantly get a high collision percentage on
        multiple [Refresh] operations. What I have seen on our net is an 
        occassional high value and many more low or 0% value for Collision %.
        Averaged out the Collision % value becomes less concerning.
    
    Bob
2425.2Misleading at the low end??SNOC02::BOYLEBRENDANBrendan BoyleTue Jun 27 1995 03:4914
    Bob,
    
    	So what you are suggesting is that given the same environment with
    	higher utilisation, this reading would become much more consistent
    	and consistently low as the increase in total traffic (denominator)
    	reduces the effect of just a few collisions??
    
    	I will check the port level stats to see the proportion of
    	collisions to total traffic...
    
    	
    Thanks,
    
    Brendan.
2425.3NETCAD::DRAGONTue Jun 27 1995 08:4420
    
    Brendan,
    
    	If the frames are pretty normally distributed and the number of
        stations constant, I'd expect lower collision percentages to be 
        reported by the Port Summary View as utilization increases from 
        your low starting point. I believe that in the higher end of the 
        utilization curve collision percentages reported by the Port 
        Summary View will start to increase.
    
        As utilization increases I'd expect to see less 0% collision
        "[Refresh] samples". 
    
        On the Port Summary View, if you use the Rate vs Percentage Option,
        this will give you a feel for just how many collisions vs transmits
        you are dealing with. Taking multiple samples will give you a better 
        feel for the traffic.
    
    Regards,
    Bob