Nagios Support Forum

Hello,

Can you explain what the representation of Bytes/Sec is comprised of, on the Summary screen?

I ask, because as I compare NNA's Bytes/Sec to my router's bandwidth screen, and my Nagios XI server's bandwidth screens, they don't seem to jive. So I wanted to see if it's a different representation of some sorts, a combination of fields, a combination of data flow over the 5 minute interval, or what.

Bandwidth in XI is only reported when the check happens. So you essentially get a sample of the bandwidth usage every check - after a day, these values will start to get averaged together as they are stored in an rrd. NNA will see ALL the traffic, so its report of bandwidth should be much more exact.

OK, I understand that, but I'm seeing what I think is a major discrepancy in the calculation or maybe my interpretation.

I have added two graphs to this reply. Once it the graph from Nagios listing is Mbps, and the other is NNA listing as Bytes/Sec. Would these two items equate in your opinion? They are from the same device. I'm just wondering if my flow device data presentation is accurate.

To me, they do not. And that is what I am trying to figure out. Nagios looks like 3Mbps while NAA looks like 30Mbps. Maybe I'm understanding something wrong here and about to look silly here. Ha!

I ran these numbers thru various internet calculators bits to bytes, and backwards.

I think you are misinterpreting. What I am seeing is about what I would expect to see, comparing the two graphs alone. Note that nna should be more accurate since it gets all data, whereas xi gets incremental counts, but that is beside the point for the moment. They are both Megabits per second, despite large or small M, and note how NNA does not denote a large or small B but XI does say lower b for bits(standard for network speeds instead of disk size(bytes)). The key reason I think you are misinterpreting the data is that nna in is looking at a 1 hour window of time, the XI graph is looking at ~12 hours of data. Specifically the nna timeline matches up nicely with the far right side of the XI graph where it has the spike in speeds.

Does that make more sense to you? Certainly feel free to argue the point if you think this is incorrect, we want to you be happy with it and trust the data it gives you! Also, no looking stupid, this is a perfectly valid question to ask, and I can see where some confusion can be had, it took me a bit to see the points above too!

OK, so the actual numbers correlate, and that make sense to me. But the NNA graph does say "Bytes/Sec", and a byte vs a bit are different by a factor of 8, correct? 8 bits to make a single byte? That is how I can up with @30Mbps instead of 3Mbps.... actually 29,845,544 bits....

But taking the NNA numbers as bps (as you have indicated), <number>/1024/1024 = 3.5578Mbps... which works out on the attached graphs....

In all, it would be my opinion to make the labels consistent across your product line, regardless of what you choose. In it's present form, it seems questionable to me. I know I've spent a day trying to prove my equipment and/or my calculations. So that's why I asked.

I would like to see "Total Bytes" stay in place for cumulative data transfers though...

You are correct on bits to bytes, and the math does seem sound. I didn't catch the popup in the middle of the graph before, but totally agree, the labels should be the same, unless you choose to specify differently. I will have to speak with our devs and verify which it is supposed to be, cause according to both our calculations, and normal networking standards it would be bits not bytes. But then the total bytes statement, I would also agree with, as it is more of an at rest disk style counter, and is generally also the standard they use. Why can't they all just get along and use one counter across it all! Hey like I said, no stupid questions, even if you just glanced at it and wondered, we are happy to figure it out, in this case I would definitely consider it a possible mislabeling.

Great! I'll look forward to seeing war comes out! Appreciate the response!

OK, so I have a definitive answer from the devs, or at least it seems to be in our minds. Both are actually in bytes. XI\mrtg is in bytes due to snmp using octets and therefore all counters being a byte, and nna being bytes, as that is what we chose to use rather than bits(they end up with funny scenarios like we have had today). So from our current understanding, your traffic is ~3.5MBps or ~30mbps depending on how you would like to count it. However both graphs still do match up, just counting in bytes, as you might have originally thought. Sorry for the confusion there, it seems kind of odd that we are counting in bytes, but that is just how the data is presented. I will put in an internal bug report to fix the XI labels to MBps as it should be.

That impossible. My internet connection is not that fast. I can't move 30MBytes/sec of data across a 6Mbps connection. That is saying I can move 240Mbps across a 6Mbps connection.

on a 100Mbps network you only get about 8-10MB/sec transfer to a NAS device. a 1Gbps network will only achieve 30-40MB/sec without jumbo frames enabled typically.

So I'm going to sit back and digest what they told you, and do a bit of research. This just doesn't add up. Maybe a label swap will make more sense once I see it in real data.

Interesting, I can speak to the devs again, but all of our research into mrtg and what data switches and routers provide, stats that it should be in bytes\octets and not in bits. They also did state that NNA was specifically written to interpret things in bytes not bits, both of the devs I spoke with are the co-developers of NNA.

I will say this though, I run NNA and XI at home, which is a regular comcast 15mbps(they state MBps, but lets face the facts here ) stated speed, I regularly see 3-3.5MBps connections from servers that are close, such as steam, orgin, netflix(sometimes) and UofM servers. I do completely agree that these are not speeds that should be within my stated connection, but it is definitely not unheard of for ISPs to untrottle connections they know will be of larger data size. Normally I see only a 1-1.5MBps connection for other services that I connect to for continual lengths of time. Getting back to the point, I checked out my system, and both NNA and XI agree with the data I see, sometimes being well above my stated speeds. I might also note that my internal network is still 10/100 not 10/100/1000.