This discussion has been locked. The information referenced herein may be inaccurate due to age, software updates, or external references.

You can no longer post new replies to this discussion. If you have a similar question you can start a new discussion in this forum.

Availability Reports incaccurate

curtom over 13 years ago

I recently was asked by management to deliver a yearly availability report. After looking at the report it was clear that the numbers were inaccurate. Almost every server showed 100% up time when each of our servers is re-booted every month. That would be impossible to have a 100% accuracy rate unless those reboots were completed within the 120 second polling interval.

I am running NPM 9.5

We are also using the Availability reports that came with the product and only modified it so that we see the production servers needed. This is a major problem we need to see resolved.

Thanks

0 mcbridea over 13 years ago

This may just be a math issue. If the servers are seen down for 6 minutes during the reboot (this is a bit generous, most will reboot faster) Then the down time for the month will come out to .013% which rounds up to 100%. to have a 99% or less uptime the servers would have to be down 44 minutes a month, or several hours a year.
Cancel
Vote Up 0 Vote Down

Cancel
0 byrona over 13 years ago in reply to mcbridea

Andy
When talking uptime or downtime a much more sensitive calculation is expected in the industry, being able to see if you meet the "five 9's" in a given year for example.
Is this grainular level of availability calculations not possible in Orion?
This is a good document to see what kind of measurements are necessary... http://en.wikipedia.org/wiki/High_availability
Cancel
Vote Up 0 Vote Down

Cancel
0 mcbridea over 13 years ago in reply to byrona

Hi Byron,
I hear people talk about five nines but normally when you demonstrate what downtime this allow (26 seconds/month) the person backs off. I know that some carriers offer five nines but they weasel it by making it a global average, not to hard to meet when you have tens of thousands of devices. The comparison I make is at the 3 nines level, I believe a much more realistic level considering current vendor equipment SLAs and industry standard polling times.
Andy
Cancel
Vote Up 0 Vote Down

Cancel
0 byrona over 13 years ago in reply to mcbridea

Oh, I totally understand how the five nines is handled in reality.
I just want to make sure I know what I can and can't measure in terms of uptime/downtime with Orion.
My understanding based on this thread is that the true five nines is not possible but three nines is?
Cancel
Vote Up 0 Vote Down

Cancel
0 mcbridea over 13 years ago in reply to byrona

Yes, but I believe it is more an industry standard rather than an Orion thing.
Cancel
Vote Up 0 Vote Down

Cancel
0 byrona over 13 years ago in reply to mcbridea

Andy
Thanks for the clarification, so long as long as I understand what the data represents, that's the important part. = )
Cancel
Vote Up 0 Vote Down

Cancel
0 Questionario over 13 years ago in reply to mcbridea

wouldnt it be easy on your part to include the ability to show a finer granularity and show like 2-3digits after the comma? I can imagine that some managers ask for it...
Cancel
Vote Up 0 Vote Down

Cancel
0 byrona over 13 years ago in reply to Questionario

wouldnt it be easy on your part to include the ability to show a finer granularity and show like 2-3digits after the comma? I can imagine that some managers ask for it...

Yeah, that is kinda what I was hoping for.
Cancel
Vote Up 0 Vote Down

Cancel