cancel
Showing results for 
Search instead for 
Did you mean: 
Create Post

GPS For Your Network

Level 11

I remember a dark time in my life when I didn't know where I was going. I scrambled to find direction but I couldn't understand the way forward. It was like I was lost. Then, that magic moment came. I found the path to my destination. All thanks to GPS.

It's hard to imagine the time before we had satellite navigation systems and very accurate maps that could pinpoint our location. We've come to rely on GPS and the apps that use it quite a bit to find out where we need to go. Gone are the huge road atlases. Replacing them are smart phone and GPS receivers that are worlds better than the paper of yesteryear.

But even GPS has limitations. It can tell you where you are and where you need to be. It can even tell you the best way to get there based on algorithms that find the fastest route. But what if that fastest route isn't so fast any more? Things like road construction and traffic conditions can make the eight-lane super highway slower than a one-lane country road. GPS is infinitely more useful when it is updated with fresh information about the best route to a destination for a given point in time.

Let's use GPS as a metaphor for your network. You likely have very accurate maps of traffic flows inside your network. You can tell which path traffic is going to take at a given time. You can even plan for failure of a primary link. But how do you know that something like this occurred? Can you tell at a moment's notice that something isn't right and you need to take action? Can you figure it out before your users come calling to find out why everything is running slow?

How about the traffic conditions outside your local or data center network? What happens when the links to your branch offices are running suboptimally? Would you know what to say to your provider to get that link running again? Could you draw a bullseye on a map to say this particular node is the problem? That's the kind of information that service providers will bend over backwards to get from you to help meet their SLAs.

This is the kind of solution that we need. We need visibility into the network and how it's behaving instantly. We need to know where the issues are before they become real problems. We need to know how to keep things running smoothly for everyone so every trip down the network is as pleasant as an afternoon trip down the highway.

If you read through this entire post nodding your head and wanting a solution just like this, stayed tuned. My GPS tells me your destination is right around the corner.

16 Comments

I built a manual solution in NPM to display and alert when any resilient links go down, and I called its view "Critical Interfaces."  It was tedious to set up, but all it involved was identifying and selecting all resilient links and putting them into a special group that is polled more frequently and which alerts more aggressively than other ports.

Interfaces I selected include the following examples:

  • Interfaces that are members of Port-channels
  • All data center access-switch ports that use LACP (for HA to servers)
  • All access switch uplinks
  • All Active distribution and core interfaces

The group is featured prominently on my NPM Summary page, access to which is given to my team of Network Analysts, our Data Center Support Team, and our Help Desk.

It's not particularly graceful to build and maintain manually, but it IS particularly useful to know when an Access Switch or a server has lost one of its resilient links.  It helps us stay proactive and reduce downtime considerably, instead of waiting for the second link to a switch or server to fail and then discovering the device is unreachable.

This doesn't cover the rest of the GPS ideas presented above, but it was a good start for me and my organization, and has proven very helpful.   My team can tell at a moment's notice that something isn't right and we need to take action.  We figure it out and fix it before our users come calling to find out why everything is running slow.

If SW has a new solution on the horizon (or even an old solution I haven't yet tried) that can better automate creating a Critical Interfaces equivalent, and automatically adding all LACP and port-channeled interfaces to it, I'm all for learning how to implement it.

When it comes to links running to our remote sites running sub-optimally, we have enough inappropriate applications running on small WAN pipes that we could benefit from something that indicated whether the throughput is problematic due to an application like Dragon (medical speech-to-text converter for transcription) or due to one of four bundled T1's failing.  It seems NetPath could be part of this kind of solution, but it does have limitations for quantities of Paths/Probes.

I have four pollers, and it appears I'm limited to 98 NetPaths by the size/license solution of my pollers.  I have more than 98 network rooms--actually, about 300 of them.

What's that you're hinting about, Tom Hollingsworth?  You've got my interest.

MVP
MVP

okay, I am intrigued....looking forward to reading about your solution.

Level 12

Ok, you have my attention.

MVP
MVP

Sounds like an ad for NPM 12 and Netpath!!  😉

Level 14

This  sounds good.  Is something around the corner?

BTW...I still have an atlas in each car and cue sheets for my bike.

Level 14

Yep.  And I like them both

Level 14

OK.  I am interested now

We currently have a 39-location WAN. Our Corp datacenter has Public and Private SONET Ring connections and two internet connections from separate providers. All connections are on divergent path. The remaining locations are tiered as: Important or Remote. All Important locations have a managed Verizon MPLS, VzW wireless backup (in the router), and then a 3rd-party SMB internet connection. Remote just has the Primary circuit and the 3rd-party internet. We've been playing with OSPF and IP/SLA for a while now to establish our own Network GPS.

   At times we feel the administrative overhead can be more trouble than its worth. But we do need to get better at this stuff.

Level 13

Been look for this.........

Level 12

Currently we use PRTG for monitoring our network equipment as well as the servers. All switch/router/core interfaces are being monitored. If one goes down, it goes red and we know about it. I have the alerts being pushed directly to my phone so I know about it almost instantly. This allows us to know of a problem, and possibly have it fixed, before the users even knew something was wrong.

I am working on expanding our use of monitoring with PRTG and adding more value to it as well, its been a fun learning process. We use PRTG because that's what was here when I took this position at the start of the year. I inherited quite a few tools that were not being utilized to their full extent, and some that were not being utilized at all, and even one that was being utilized improperly.

If you're not using Orion yet, and if you get a chance, download a trial version of NPM and NCM and install them so you can compare functionality.  I grew up on MRTG, then HP OpenView, then Nortel Enterprise Switch Manager, and finally the SW Orion suite.  This one does it all for me, and has many great modules to add on for specialty knowledge discovery and alerting.

It's a wonderful thing to have all your data in a single pane of glass, and to be able to diagnose latency issues immediately to their cause, whether bandwidth congestion across a WAN (AND to be able to find the specific cause and application and IP address--through Netflow Traffic Analysis with NPM), or to see a SQL server is doing latches in two seconds instead of two microseconds, or whether someone has made a configuration change to a switch that's causing the issue.

I can't say enough positive things about Solarwinds products.  They save my company more money every year, by decreasing our down time through prevention and discovery and analysis.  And they give us an edge for predicting growth, which keeps us predictive instead of simply proactive or reactive.

Level 20

I have ntp server device that runs off gps signals!

Seriously?  It wouldn't surprise me, but if you're not joking, please tell us more about how you integrate GPS with NPM.

Or did my leg just come off in your hand?

Level 20

It's how we get an accurate time source for ntp without any wires to the outside world... we get the time directly from the GPS satellites... just a little antenna.  Common with important air gapped networks.

Ah, yes--we have that, too.  Installed the antenna up on the roof back in 2004.  The joke was that, since this NTP is GPS-based, we'll not only have our timing correct, but we'll also know when our building moves, and where it's gone to.

pastedImage_0.png

Level 21

While it's certainly an Ad for NPM 12 and NetPath, it's certainly worth of an Ad as it's technology in an NMS unlike anything I have seen before.  I am really excited to see that SolarWinds is back to innovating as it feels like it had been a while since I had seen them produce any new tech that I had been excited about.

About the Author
A nerd that happens to live and breathe networking of all kinds. Also known to dip into voice, security, wireless, and servers from time to time. Warning - snark abounds.