How is % Packet Loss calculated?

Hi,

packet loss is calculated as a moving average of 10 last ICMP echos. Each ICMP request has value SUCCESS/FAILURE. Those results are store into buffer with length 10, and we calculate average on it.

Example:

9 success and 1 failure ping = 100 * 1 / 10 = 90% packet loss

In case we have in memory only 2 pings, because we restarted all services

1 success and 1 failure ping = 100 * 1 / 2 = 50% packet loss

We send ICMP pings according your polling interval and that's 120 seconds by default. But there's also in play "Fast Polling".

Hope this help

thanks for the reply.  Now I have good understanding of % packet loss.

But in my case, Orion app was not restarted. my chart is set to show every minute.  my device polling is also every minute. the minute prior to my 50% packet loss, it shows 0% packet loss, the minute after, it shows 0%.  Based on your description of packet loss calculation, shouldn't I see a gradual increase then gradual decrease?  What do you think happened?

Oh, you mean chart and statistics data. I was talking about information on Nodes Detail resource. Chart and statistics is a bit different. We store just raw data according your polling interval.

0   ........packet loss for successful echo

100....... packet loss for failure echo

Each minute we store just 0 or 100 based on polling interval. Then for instance on web resource you query last hour, and chart sampling interval is 5 minutes. We compute average for all pings happened during each 5 minutes.

So if your chart sampling is 2 minutes, and just 1 ping failed, you can see packet loss 50%.

Is it clear now?

This this not be 10% and not 90%? I would expect 90% to happen with one success and 9 failures?

9 success and 1 failure ping = 100 * 1 / 10 = 90% packet loss

Sorry for confusion. Of course it behaves as you described

9 success and 1 failure ping = 90% availability and 10% packet loss.

but sometimes, i could see 6%, where is it from if the base is 10 packets ?

Hi All,

What do you mean in two line

1. 9 success and 1 failure ping = 100 * 1 / 10 = 90% packet loss

2. 9 success and 1 failure ping = 90% availability and 10% packet loss.

I think this is correct:

9 success and 1 failure ping = 100 * 9 / 10 = 90% availability