Eval NPM version 9.5 SP4

Question

Hi Everyone,

My company is interested in purchasing this product to monitor our production environment.  The production environment consists of about 400 Cisco routers and switches, 10 Cisco load-balancers CSS and ACE, about 30 Checkpoint NGx R65 firewalls running on Secureplatform, 10 Riverbed Steelhead appliances, 40 Cisco wireless APs and 2 Cisco wireless controllers, about 300 virtualize Windows servers, 200 Solaris 9/10 SPARC and about 100 Linux servers (Gentoo, CentOS, Redhat Linux ES 3/4/5 & Unbuntu).

As a pilot, I installed NPM version 9.5 on a DELL 2950 dual-processors with quad-cores and 12GB of RAM on a RAID-5 HDD.  The machine is running Windows 2003 with SP2 and up-to-date Microsoft patches.  After installing NPM version 9.5, I installed SP4 on top of it and reboot. Currently, I only monitor 3 linux servers, 3 Windows 2003 Enterprise Server, a Gentoo Linux server, a CentOS server, a Checkpoint Secureplatform firewall and a couple of Cisco catalyst 3550s switches.  In other words, the NPM box should be handling these devices without any issues.  However, here are some of the issues I have about the product:

#1:  When monitoring both Windows servers, Linux server and Checkpoint firewall, I noticed that the bandwidth utilization is NOT showing up correctly.  When the interface is set to 100Mb, it shows up in system manager as 1Gb.   The bandwidth showed that the 100Mb interface is pulling about 190Mbps.  This test is repeatable on various Windows, Linux and Checkpoint firewalls.  The bandwidth graph, however, looks correct on Cisco switches.

#2:  When pulling up the web interface from NPM host via web console, http://127.0.0.1:8787, it is EXTREMELY slow, very frustrating for me and my colleagues.  This makes managing/editing objects extremely UNSCALABLE.  For example, if I want to change to counter to 64-bits, this has to be done from the web console.  This option was available in version 8.5, 8.5.1 in system manager but I can no longer do it in system manager in version 9.5.

#3:  I am setting basic alerts on on interfaces of my Linux servers to send me email alerts whenever the server interface transmits or receives more than 25Mbps.  I got email alright but I am getting email alerts even when the server transmits or receive 4Mbps.  Something is definitely wrong here.

The network consultant working for us keeps telling me that Solarwind NPM is a very good product;  However, he can not address these issues I described above.  These issues really concerns me.  I do not want to purchase this product until these issues are resolved.

Anyone with NPM 9.5 SP4 running in production environment is experiencing similar behavior that I am seeing in my eval environment?

Thanks in advance.

FormerMember · Answer

Hmm.  Well, there goes my theory.  I thought you were monitoring more than that.  A single server should be enough if you're not a VM.  Those are just guesses based on typical problems.  From your description, it should be performing well, and it's tough to troubleshoot odd problems on a forum, so the SE is probably your best bet.

netmonguy · Answer

"You should probably get a Sales Engineer involved to take a look at the bandwidth issue.  That sounds like a configuration problem."

There is a Sales Engineer working on this.  Will see how it goes.

"The slowness seems straightforward.  Even in your eval, you are monitoring a lot of interfaces.  But you are running on SQL Express on the same server.  SQL Express is limited to 1 CPU and 1GB of RAM.  Try a full version of SQL on a separate box."

Not sure what you mean by this.  I only manage a few servers, firewalls and switches.  The server is running at less than 10% CPU utilization.  I was told by YOUR SOLARWIND SE that a single server is sufficient for my environment.  We are currently running Open Source NMIS on the same hardware and monitoring about 900 devices without any issues.  If we have to use more than a single server, it will be an issue.

"And if you're running that SQL Express on a VM, it'll be even worse.  SQL doesn't perform well on a VM without very specific configuration."

We do NOT run Solarwind or any kind of monitoring systems software on VMWare, only on high-end servers with fast CPUs/quad-cores and LOT of memory.

FormerMember · Answer

You should probably get a Sales Engineer involved to take a look at the bandwidth issue.  That sounds like a configuration problem.

The slowness seems straightforward.  Even in your eval, you are monitoring a lot of interfaces.  But you are running on SQL Express on the same server.  SQL Express is limited to 1 CPU and 1GB of RAM.  Try a full version of SQL on a separate box.

And if you're running that SQL Express on a VM, it'll be even worse.  SQL doesn't perform well on a VM without very specific configuration.