NPM SL100

We purchased the NPM SL100 to monitor our Juniper ex3300-48p switches. There is about 30 switches total. There are 10 stacks. However when I run through the setup I can only monitor a very small part of the switch information. The representative told me I could monitor all the information I needed from the SL100. Any ideas on best things to monitor on switches? If I put all content then it shows 941 over the limit of 100.

Find more posts tagged with

interface monitoring

npm sl100

Accepted answers

All comments

mesverrum

the 100 license will monitor a max of 100 network interfaces, presumably on 30 switches that might be keep an eye on just the uplinks between them and not much else.

rschroeder

The SL100 lets you monitor 100 "elements" (100 interfaces, 100 volumes, and 100 nodes). It may not be big enough to let you monitor all interfaces on even a single node. This is because a node may have many interfaces. If you want or need to monitor more than 100 Interfaces, there are things you're going to have to prioritize. What interfaces to monitor, what to let go. Fortunately, NPM will get you a LOT of information by monitoring a node with SNMP.

I'm a Network Analyst; the vast majority of my Solarwinds elements point at interfaces, but I'm also concerned with seeing all the monitorable parts of switches and routers and firewalls. My SysAdmins take care of the servers, so I monitor those devices' volumes out of interest, and to provide a reality check against the SysAdmins' monitoring of the volumes with their own tools.

An element is something Solarwinds monitors. One of my nodes can contain MANY elements (especially interfaces)--and you may wish to monitor some--or ALL of them. Or you might choose to monitor none of the interfaces (or only 100), and just rely on ICMP response for some nodes--ones which you're content to simply know if they're up or down, when you don't care about their internal stats or temperatures or volume sizes, etc. But it's nice to know you can monitor all of your switches as "nodes" AND still have room to monitor servers and UPS's and other devices as "nodes" too.

Some of my chassis switches (which are each "nodes") might have 500 elements within them--one for every physical port (384 per chassis). But the "node" monitoring for them covers the snmp-discovered temperature sensors, CPU's, power supplies, fans, and line cards--but not disk volumes. Once you start monitoring volumes, the count begins over at zero until you've monitored 100 of them--the size of your particular license. The same can be said for a stack of switches, or for a server's resources--each one is a node, each node can contain many elements that are interfaces, volumes or nodes, and the count of elements you can monitor is what the license is based on.

And you can prove that all of these resources are elements (not contained under one "node" for licensing purposes) by simply watching your poller's statistics as you add or remove interfaces, nodes, or volumes:

If your budget constrains you to a license for only 100 elements, consider the following:

Use snmp only to monitor each of your 30 switches. It consumes 30 elements of your SL100 license--one per each switch. This at least lets you know they're up, and when they go down. But it won't tell you why they went down, or give you data for trending and forecasting future growth needs of Internet or WAN pipes or database sizes or volume sizes or memory, etc. That's where the 100 license count gets to be used for Interfaces.
So you have 70 "node" elements left to use up. Monitor UPS's, servers, firewalls, routers, load balancers--anything that can be pinged and discovered with snmp. You may even have left over unused "node" licenses--I do!
Then think about which 100 Interfaces to monitor. Perhaps monitor the uplinks between devices, or network ports of high interest, or Internet or WAN bandwidth. Maybe ports going to servers. There are MANY things you can choose to monitor. Figure out what your biggest cause of outages and Help Desk tickets is, and use your remaining Interface element count to monitor those indicators of possible trouble, and set up your system to let you know BEFORE an interface gets into trouble with many errors or runs out of bandwidth.
Last is your 100 volumes. Certainly your servers have volumes. So do firewalls, load balancers, and even new switches do. Or, maybe you'll monitor volumes in PC's. It's yours to decide how to use that 100 count best.

Don't forget to make this information available to the right people. If you are the only one with access to the info, everyone's going to keep calling you for it and you'll still be in a reactive environment that no one wants. Make the info easily available to your boss, to your Help Desk, to your Server Admins, etc. Let them see for themselves that it IS or ISN'T the network. And you'll be on your way to a predictive environment, and then to a proactive one, and finally to the place where you are able to confidently offer SLA's to your customers. Without sacrificing your home life.

When you discover you really need insight into more than 100 elements (when it comes to monitoring interfaces, this is a "when", not an "if"), learn ways to show Finance and Management that Solarwinds isn't a cost center or liability, but that it's an asset because it will help keep your business running well. Show how the right sized licenses and Solarwinds modules will reduce business operating costs by increasing up time through reducing troubleshooting and reaction time. Show how these tools enable your teams to discover a problem and troubleshoot it MUCH more quickly. We call this "MTTI": Mean Time To Innocence. NPM really helped me reduce the time it takes to KNOW that "It's not the network" and to send the ticket on to the right team for analysis and troubleshooting.

I recently switched over from four pollers with "Unlimited" SLX licenses (for 12,000 elements each) to the NAM licenses. NAM got me four new Solarwinds modules (IPAM, UDT, VNQM, and HA) in addition to the NPM/NCM/NTA I already owned, and reduced my annual support contract cost. At the same time it let me increase the base of customers that rely on Solarwinds via the new modules. It was a win-win from my point of view. Through NAM I can monitor up to 100,000 interfaces, 100,000 volumes, and 100,00 nodes. With 100 hospitals and clinics in my network, with close to 300 network rooms, monitoring every switch port and node is important. That 100,000 element top end for Interfaces is not wasted money

Funding is often more about impressions and spin than you'd expect. Ensuring people get the right impression will help your projects be properly funded.

If you're interested in learning more about how to get money for bigger license counts, I shared my thoughts about how to succeed in that endeavor here:

https://thwack.solarwinds.com/community/solarwinds-community/geek-speak_tht/blog/2015/09/08/a-strategem-for-obtaining-fu…

cjfranca

You can to monitor 300 elements. 100 cpu 100 volumes and 100 interfaces

Enviado do meu iPhone

Em 4 de nov de 2017, à(s) 17:17, rschroeder escreveu:

NPM SL100
reply from rschroeder in Network Performance Monitor - View the full discussion
I'm not certain you got the right information. The SL100 lets you monitor 100 "elements", not 100 nodes or devices.

Some of my chassis switches have 500 elements within them~~one for every physical port, another for every temperature sensor, more for CPU's and power supplies and fans and blades. The same can be said for a stack of switches, or for a server's resources~~each one is a node, each node can contain many elements, and the count of elements you monitor is what the license is based on.

If your budget constrains you to a license for only 100 elements, consider the following:
Use ICMP only to ping each of your devices. This at least lets you know they're up.
If you have enough elements left over, then perhaps monitor the uplinks between devices, or network ports of high interest.

When you discover you really need insight into more than 100 elements, discover ways to show Finance and Management that Solarwinds isn't a cost center or liability, but that it's an asset to help keep your business running well. And show how the right element licenses and modules will reduce costs by increasing up time through reducing time to discover a problem and troubleshoot it.

Funding is often more about impressions and spin. Ensuring people get the right impression will help your projects get the right funding.

If you're interested in learning more about how to get money for bigger license counts, I shared my thoughts about how to succeed in that endeavor here:
thwack.solarwinds.com/.../a-strategem-for-obtaining-fu…

I recently switched over from four "Unlimited" SLX licenses for 12,000 elements each to the NAM licensing/products. I got four new modules and I reduced my annual support costs while increasing the base of customers internal to us that rely on Solarwinds. It was a win-win from my point of view. And now I can monitor up to 100,000 elements, and believe me, with 100 hospitals and clinics in my network, with close to 300 network rooms, monitoring every switch port and node is important, and that 100,000 element top end is not wasted money.
Reply to this message by replying to this email, or go to the message on THWACK
Start a new discussion in Network Performance Monitor by email or at THWACK
Following Network Performance Monitor in these streams: Email Watches, Inbox, Following
Following rschroeder in these streams: Following

alexch

SL100 = 100 network objects in summary. It means you can monitor 100 nodes, interfaces in total (10 nodes and 90 interfaces for example). 1 stack of switches = 1 object.

mesverrum

Incorrect, the license model is a little confusing but the number in your npm license is not the total elements allowed, it is the max of nodes, interfaces, and disk volumes. Not in total but in each category. So you can monitor up to 100 nodes, and up to 100 interfaces, and up to 100 volumes. Since the OP specifically mentioned switch stacks the most likely limit they will encounter is the 100 interfaces, but even after that limit is hit they could still continue to add nodes until the total node count was 100. Things like monitored cpu's and hardware health sensors and universal pollers do not count against the license limits in any way. I would also point out that since these are juniper switches they will reach have a bunch of disk partitions available to monitor but in 99% of cases you don't really need to add most of them, if you even bother to monitor storage at all on the switches.

cjfranca

100 nodes= 100 CPU

jeff.stewart

The SL100 allows you to monitor 100 nodes, 100 volumes and 100 interfaces. This document should help; NPM licensing levels

Jeff

brianj

Not accurate, at least not for NPM. A server with 16 quad-core CPUs is one node.

grovernz

Does that include Hyperthreading?

brianj

I got my information directly from SolarWinds. See https://support.solarwinds.com/Success_Center/Network_Performance_Monitor_(NPM)/NPM_Administrator_Guide/Licensing_SolarWinds_Network_Performance_Monitor/NPM_licensing_levels

This page states the SL100 license covers "Up to 100 nodes, 100 interfaces, and 100 volumes (300 elements in total)."

rschroeder

Right. Believe that--it's the right information.

In my environment, it happens that Interfaces make up the vast majority of my elements, so my personal focus is on them. I manage fewers servers, so I tend to ignore the volume licensing since it's automatically covered in my SLX "unlimited" licenses. The same for my node count, and I apologize for any confusion I may have created earlier in that regard.

You're right--you can manage 300 elements--up to100 of any three: Nodes, Interfaces, and Volumes. You can't manage 40 nodes and 260 Interfaces. Although that might add up to 300, you're only allowed up to 100 of any one category.

I manage 849 nodes (which covers snmp monitoring of the internal basics, but not the nodes' interfaces, nor any volumes on nodes).

My five polling engines show this amount of nodes:

I ALSO manage 42,746 interfaces on those 849 nodes, which covers physical ports and virtual ports, along with 491 volumes.

I initially started with an SL500 and quickly discovered one or two full-loaded ten-slot chassis switches ate that license up with their interfaces. Sure, I still had room for more "nodes" and more "volumes", but I couldn't manage all of my Interfaces, and that's important to my organization. So I did a quick estimate of how many devices I had and I upgraded to three SLX "unlimited" licenses (12,000 interfaces is the limit per SLX license, so it's not truly "unlimited" to my way of thinking). Eventually I got the EOC, but didn't like having three separate companies showing up in it when my three regional organizations are all under one roof, all one company. So I dropped the EOC and moved to NAM, which got me the ability to monitor 100,000 interfaces with more than 20 pollers. Doing so enabled me to reduce my annual support costs due to Solarwinds fee structure. And I was able to add on VNQM, IPAM, UDT, and HA along with all those extra Interfaces.

Now if I could get the right silos to be interested in VMAN, SAM, WPM, and DPA, I'd be set for the next few years. Ha!

Determine what you want/need to monitor, then decide what you can monitor (nodes, volumes, and interfaces) with the licenses you can afford. You'll do fine, and you'll make decisions about whether you want/need to budget for more licenses in your next fiscal cycle. Maybe you won't need additional licensing. Maybe you need a little more, or a lot more, to do the job the way you feel is right.

The important part is getting your job done well, within budget, and reducing down time. And doing it so that your peers and your boss can benefit from it.

Which can make you the company's rock star of the year!

rschroeder

Thank you for clarifying, Claudia. I've edited my original response to be more in line with your observation.

Good job!

sja

Hi tnewton

Well if you can not get the right license...

I will do that

Set all the switchs in that SL100 (30 nodes)

Set all The uplinks and the VC interfaces (up to 100 )

Set UNiversal poller on all the nodes to get juniper red/yellow alerts (that is free)

Set the oob hw monitor to get temp/power..(that is free)

now if there are still interfaces that are important

you still have syslog and trap to get the up/down status (that free)