Comments on: Does anyone measure the uptime or downtime of servers?

By: jim4522

jim4522 — Fri, 08 Oct 2010 16:47:34 +0000

Carlosdl, thanks for your response. It is not clear to me what the purpose these uptime numbers or percentages would serve. They are not indicative of the reliability of the server since uptime and down time are more a factor of the maintenance option the organization has chosen. A better indicator of reliability if you are talking about all the things that can cause a server to be shut down is either the number of shut downs or reboots (they should be almost the same if not the same). And as a CIO if I were intertested in tracking compliance with my SL agreements the number I really want is the uptime of the application not the hardware. End users could care less about how long the hardware stays down as long as my application failed over almost immediately to a clustered backup. The other problem that I see with uptime data is that it is not predictive, it goes up and down each month giving no clue to the real stability of the network that the servers are serving. Uptime data is not actionable. It doesn’t change what IT does and more imporatantly it doesn’t change what vendor do. Thoughts? Jim4522

By: carlosdl

carlosdl — Fri, 08 Oct 2010 13:17:07 +0000

Most monitoring solutions I have evaluated (zenoss, zabbix, nagios, among others) provide some kind of availability or uptime report (usually expressed in percentages).

By: jim4522

jim4522 — Fri, 08 Oct 2010 13:03:34 +0000

Jim Teik I appreciate you response it is helpful. Jim4522

By: jinteik

jinteik — Fri, 08 Oct 2010 02:32:17 +0000

Hi Jim,

I will answer what I can again (following the way my office does it)

1. Uptime means that the OS is ok and that it is accessible by everyone.

2. This question I cannot answer you…as i have no answer for it..but if there is a need to do reboot we will inform the people who are accessing it 1st…

3.Actually at our side, the management report will show everything like incidents(if the system went offline, show the report number and etc), plan down time for updates of patches and etc (this is to be highlight to user in advance before carrying out this process)

4.Not sure of a software coz we do it manually.

By: jim4522

jim4522 — Thu, 07 Oct 2010 22:25:14 +0000

Jim Teik responded, “In data center they do because they need to give report of system up time every end / beginning of the month.

I have four follow-on questions to Jim Teik’s answer:

1. Is the uptime that is reported the uptime of the hardware or the operating system?

2. If I have a virtualized server that is running 10 instances of the operating system and the hardware runs exactly one month between reboots, do I log one month of uptime because the hardware was up for only one month, or do I log ten months of operating system uptime because each of the operating systems were rebooted?

3. In this monthly report to management do I also report the number of reboots that occurred each month so that management knows how many incidences occurred?

4. Is there existing software does all these accounts all these counts?