Hello,
What solutions do you guys in the fixed network business/ISPs use to provide customer portals for network KPI reporting to customers in a fixed network on real time basis. The KPI in question are network availability, utilization, memory/cpu of managed routers/firewall, jitter, packet loss etc in a multi vendor environment.
What would you recommend especially in the licensed/supported options and not the free ones like Zabbix, Cacti, MRTG etc. This solution should scale well for hundreds of thousand of clients.
We have been using Orion NPM and it pretty much does the job but would wish to move to something more scalable for SP environment.
Cacti and MRTG are distributed under the same license.
Zabbix, at least, is supported commercially by a company consisting of
grown-ups who know what they're doing. They can get their platform to
do really cool things and to scale better than you could do, because
they know the platform better than anybody else. I work for another
company that has a similar relationship to a different free-software
management platform; see the domain part of my e-mail address if you
care which one it is.
Whichever platform you choose, you should be prepared to spend far
more on consulting (or pricey headcount) than on acquisition or
maintenance of hardware and software. For any platform to scale to
the extent you're talking about, it's going to require extensive
expert customization as well as constant care and feeding. I've been
managing networks in-house and in consulting roles for over a decade,
and this rule always applies.
My point in all this is that, while most free-software platforms suck
at scaling up because they're not designed for huge scale, you should
not dismiss this entire class of platforms out of hand any more than
you would dismiss all proprietary offerings on account of Orion NPM
missing the mark.
I'm getting ready to do an eval of Monolith Software's monitoring/management product. They have some very nice multi-tenant dashboarding and reporting capabilities and are extremely scalable.
We've just deployed Intermapper to do all of our device polling, link status and topology mapping. Works very well and looks real pretty.
For graphing, we use cacti with the Discovery and Autom8 plugins.
For SNMP trap parsing, we use SNMPTT.
We're currently evaluating Splunk to eat the SNMP trap and syslog data from our gear and do cool stuff with it.
Last on my list of tools to try is Cisco NCM as a replacement for RANCID. RANCID is amazing, but when we have hundreds of devices with exactly the same base configs on them, something a little more sophisticated than RANCID is required to keep all of those configs in sync.