search cancel

How is latency determined (Legacy KB ID CNC TS5367 )


Article ID: 52064


Updated On:


CA eHealth


By default, eHealth polls statistics elements (LAN/WAN, router, server (except NT Registry), etc) every five minutes. The steps in the polling cycle are as follows:

1.        Ping the first 100 devices in the poller configuration.
           - Response is used to determine whether the device is alive and to collect latency data.
           - If the device does not respond to this ping, another ping is tried.
           - If all ping retries fail, the device is considered unreachable and eHealth shows a Missed Poll.
2.        Send out an SNMP get for the statistics we are collecting, as determined by the .mtf file associated with each element.
3.        Send out the next 100 pings.

How the Poller utilizes ICMP echo requests (Ping)

?        Does not send ping requests on fast polls
?        At the beginning of the poll cycle, one ping per IP Address is sent probe the network
?        Experience has shown that the first ping often has a longer latency because routers and other devices in the path may have to load ARP caches or establish links.
?        Results from this ping are discarded
?        Ping requests, as many as defined in NH_POLL_PING_NUM_PINGS, are sent to each IP Address.
?        If multiple elements have the same IP address, only one set of requests is sent to that address.
?        Ping requests have a packetsize of 100 bytes, or as defined by NH_POLL_PING_PKTSIZE         
?        From the ping results, the minimum time is used for latency, or as defined in NH_POLL_PING_MEAS_TYPE

Variables related to Ping:

?        The number of requests we send out at a time each poll cycle.
?        Default value: 100.
?        Decrease to reduce the amount of load we create on a large device each poll cycle.
?       Controls the number of pings sent after the initial ping used to probe the network
?       Default value: 3
?       Defines which result is going to be taken from the pings (min, max, avg)
?       Default value: min
?       Controls the size of the ping packet we send out to each device
?       Default value: 100
Controls whether we ping devices before we sent the SNMP Get. Default is enabled. Set to disable to turn off (this will also disable the ability to collect latency statistics).

Related Issues/Questions:
latency values compared with command line ping or ping results from response path tests
How is latency measured
How is latency determined
How does eHealth calculate latency values
Why are latency values different than those seen from a command line ping

Problem Environment:

(Legacy KB ID CNC TS5367 )


Release: LHBASC99000-6.2-eHealth-Live Health Upgrade Elements