DX Unified Infrastructure Management

Friends,

I am part of an MSP that monitors many data link.

I would like to hear from those who use the UIM (Nimsoft) for this purpose on a large scale, the recommended settings.

Example:

Check in Interval, Retries Interval, Packt loss, Jitler and Lantecy.

I have done numerous tests and if the profile is not fine tuned, does not work properly.

Thank you.

Availability check is the most simple monitoring and the most difficult monitoring.

So it is important to place the probe to closest location for target to avoid false alarm from occurring by unknown factors (network disconnection, routing issue, etc...)

Hi,

Have a look at the saa_monitor probe.

Have used this probe in the past for IPSLA type monitoring. 

Thanks

Hi,

saa_monitor is a good tip. But works just with Cisco devices, doesn't it?

Hi Ronil,

You could use snmpcollector or snmpget with the vendor OIDS to gather the similar information. I understand that most vendors would publish this information on their websites. 

Thanks,

Mark

Hi friends,

what is happening in my environment that is next ...

I have a robot with the probe of net_connect that pings tests to 216 profiles.

The check interval is 60 seconds.

CA Support says the probe of net_connect was developed and tested for a much larger number of profiles than I have.

I've done numerous settings on the probe to avoid ( "Profiles, failed to perform in schedule interval time") without success.

The best scenario I found was using the robot on a linux server. Where I managed to get a greater number of profiles.

I remember the same robot that use net_connect 216 for these profiles, I also use the interface_traffic / snmpget for the same 216 profiles.

I do not know if there is any correlation between the performance of probes that may be causing these delays.

I'm currently installing two robots linux to separate these 216 profiles, on order to solve this problem.

I wanted to hear from friends as you work in that kind of scenario.

Thank you.

Hi,

I found the case you raised with us in the system.

There are a couple of things to try:

1 - increase in the interval time (from 1min to 3 or 5min)
2 - Lower the Packet to Send attribute
3 - Lower the Delay between Packets(ms) attribute

It could be that expecting the probe to run 200+ profiles with these attributes set all within one min wont complete before the next expected poll cycle.

Hi mark,

I know these recommendations, but I need to work with 60 seconds interval.

3 or 5 minutes to inform you that the data link is out is long.

We need to be aggressive with this type of monitoring.

I did tests for 3 days and did not like that the probe works.

When the probe is not ping problems or packet loss, works fast.

But when she finds some delay, loses runtime.

Example:

1 - The runtime between a profile is 60s;
2 - Delay of 12 sec;
2 - It starts three times ping with 10 second intervals between one and othe ping;
3 - Delay 2 sec + 10 sec + 10 sec + 10 sec = 32 sec.
4 - Start packet loss test;
5 - Send four packages with 4 sec interval;
6-32 sec + 10 sec = 42 sec.
7 - + some delays = 50 sec

Theoretically the settings are adjusted to work well, but the composition of the tests that cause delays.

My Settings are as follows:

Check Interval: 60sec

ICMP
Packet size: Default
Retris in interval: 3
Packet to send: 4
Delay between Packet (ms): 1000
Packet Loss (x) Jitter (x) Latency (x)

Connectivity
Timeout: 10
Failed intervals: 1

Theshould
Response time: 2000

Packet Loss
Max lost packets: 2

I wanted to hear from partners who use this probe with many profiles or with short intervals, so as to reach a conclusion.

Really I have to have multiple robots to perform the profiles with short breaks?

Will use multiple robots on the same server is an option?

Thank you.

Hi,

The solution to this case was to separate a certain amount of profiles on some net_connect servers.

Thank you.