Self-hosted network monitoring
Every hop. Every minute. On your own servers.
FishEye pings, traces and polls your network from your own sites, keeps every hop of every traceroute, and with flow shows what filled the link.
- Every minute a ping check, by default
- Every 10 s if a target needs it
- 80,000 flow records a second, stored in a lab test
- 90 days of traceroutes
Start where you are
What are you trying to find out?
Is it down, or just slow?
Every target is up, down or unknown. When the probe that checks it goes quiet, it reads unknown, not down. Alert rules on loss and round trip catch the slow ones.
How FishEye answers itIs it us, or the far end?
Probes at your other sites and agents on your users’ computers show the network from more than one place.
How FishEye answers itWhere on the path?
- Every router your traffic passes on the way to the target, with its network and where it is.
- Loss and round trip for each hop, now and over time.
- Scroll back to the route before it changed.
What filled the link?
With the Flow add-on, your routers report who talked to whom and how much, so you see the hosts, networks and ports behind the traffic.
How FishEye answers itIs the hardware failing?
Your routers and switches report their own health over SNMP: traffic, errors and state on every port, the modules and serial numbers inside them, and the traps and syslog they send.
How FishEye answers itWho needs to know?
One board for the team watching the network, alerts to Telegram, email or a webhook, and a storm of alerts shown as one row.
How FishEye answers itWhy check often
A short problem is easy to miss.
Many monitoring tools check once every five minutes. A slowdown that starts and ends between two checks never shows up, even though your customers felt it.
FishEye checks each target every minute by default, with five pings each time. For a link that matters, set it to every 10 seconds, and a 40-second slowdown leaves three or four readings instead of none. Your plan never limits the interval.
Every 10 seconds the fastest setting, for any target
Four readings catch the spike.
Every 5 minutes how many tools poll
Both readings land either side. Nothing is seen.
Set up your way
Watch each link as closely as it matters.
Every check, device and alert rule has its own settings. A small office can keep the defaults; a NOC can tune each link. Same product, same plan.
Most of the network
The defaults just work
Ping each target and poll each SNMP device once a minute. Add a target and it is watched, with nothing else to set.
The links that matter
A close watch where it counts
Ping your uplinks and core every 10 seconds, and traceroute them from each probe as often as every minute, so a slow hop shows on the next run.
Alerts
Rules that fit your network
Each rule sets its own threshold, how many results in a row open it and how many close it, and whether it covers one target, one interface or all of them.
One install
Six tools. One login.
Most teams stitch a pinger, a traceroute box, an SNMP poller, a flow collector, endpoint scripts and an SLA spreadsheet together themselves. In FishEye they share one login, one set of access groups and one set of alert rules.
How it is built
Measured, not promised
Room to grow, on modest hardware.
Your routers can send FishEye a flow record for every conversation that crosses them: who talked to whom, and how much. We load-test FishEye on one small machine that also runs the traffic generators. Here is what it kept up with.
A lab test with generated traffic, not a customer network. More on capacity.
See your network the way it behaves.
FishEye is made by Layer 10. Tell us about your network, and we will walk you through it.
Talk to us