On Thu, Dec 21, 2006 at 07:14:43PM -0800, RAMA wrote:
I am evaluating Hobbit for monitoring a network infrastructure with 3000+ servers. Do you have any hobbit screen shots for such large implementations? How 'Status "at a glance" ' can be achieved with hobbit for large number of servers?
My current production system has 3800 hosts in it.
The key to making this usable is to be very specific about what you want to label as "critical". Use the "Critical Systems" page for your 24x7 operations monitoring, and split the hosts on several pages so each of your operational groups have their own hosts easily accessible.
Have your 24x7 operations people use the ack function on the critical systems view, so they will always see only the events that haven't been assigned to someone.
I know some people rely on the "All non-green" view for monitoring. That's completely useless with that number of hosts. Right now I have 25.585 individual statuses being monitored, and 2389 are yellow or red (eg lots of Windows systems' "msgs" - Event log - columns). It rarely goes below 2000.
If your network has some critical paths - ie when one router dies you lose access to a large number of hosts - then consider using the "route" tag in bb-hosts to stop the hosts behind that router from all going red when the router dies.
Use e-mail alerts for those things that are not immediately critical (eg a disk filling up), but which need attention soon. For the critical items, alert to a pager or send an SMS.
Regards, Henrik