Community
- Command Central
- MVP Program
- Monthly Mission
- Blogs
- Groups
- Events
- Media Vault
Products
- Observability
- Network Management
- Application Management
- IT Security
- IT Service Management
- System Management
- Database Management
Content Exchange
- SolarWinds Platform
- Server & Application Monitor
- Database Performance Analyzer
- Server Configuration Monitor
- Network Performance Monitor
- Network Configuration Manager
- SQL Sentry
- Web Help Desk
Free Tools & Trials
Store

True PC spec required?

We are currently evaluating Orion as an NMS, and have to say have been very impressed so far - bit of a shame about the severe limitations of the demo product though.

We would be looking at using the SLX version of the product to realistically monitor somewhere in the region of 4000 router and switch interfaces.

I assume that there must be quite a number of you out doing axactly this at the moment. What has the performance of the product been like under those sort of loads? We would be looking at having somewhere in the region of 8 monitor accounts logged in to the product at any one time.

Does anybody have any sort of comments on product performance and PC platform required to get a satisfactory performance under those kind of loads. (Looking for real world indicators rather than Solarwinds recommended spec). Thanks!

Find more posts tagged with

Accepted answers

All comments

hamoja

Orion 6 for the most part satisfies my SNMP requiements for my Cisco farm (router+switches+wireless) and 2000 servers. I am managing over 2500 interfaces and CPU utilization is inconsistently between 30%-80% . The product honestly has minor bugs but its a work in progress and support has promised to resolve these issues.

I am runnign SLX build 6.3x for 4 months now and these are the bugs I have identified:

1. Cisco switches with CATOS intermittently report interfaces as unknown or down when they are up. They are working with Cisco to resolve this very huge issue. IOS works great.
2. At least once a day IIS service (iisreset) should be restarted. Ocassionally the web site cannot be found
3. AS400/Novell servers are not properly polled. Inaccurate readings on most polls

My PC specs are: Dell 2.8GIg with 1GIG or RAM (highly recommended), Windows 2000 w/sp3+patches.

Good Luck

colinpryke

Thanks for that. Interesting point about the CatOS - we are looking to monitor somewhere in the region of 120 cat switches, so this could be a fundamentally serious bug for us! We have also been revising our element count and think that it could now be as high as 12,000. We would also like to poll these devices for up status more frequently than the default five minutes (1 minute would be ideal). Has anybody wound up the poll rate to that extent? Is it reliable? We are looking at putting a 1000baseT card in the server so network bandwidth should not be an issue. How much disc capacity should be factored in for long term stat gathering on this number of elements?

MrSNMP

The issue with monitoring switches running CatOS is that when under stress the switches start dropping SNMP requests. Version 6.4 of Orion NPM includes features to work around this issue. I haven't seen the issue reoccur since upgrading to this version. With 12,000 elements you may need to use 2 polling engines. Also, separating SQL onto a separate server boots performance significantly.--HTH

colinpryke

Ah, that should be ok then - although we have a significant number of switches they are not particularly busy - not busy enough to dump SNMP anyway! We were going to split the SQL databse onto another server with a 36gb drive array. The polling engine we were considering would be a dual 2.8gb Xeon Compaq proliant server with 2gb RAM and dual (for redundancy) 1000baseT network cards. Would that still require a second polling engine do you think?

BryanBecker

Yes..the CatOS issue is fixed in 6.4 I had the same issue and it worked afterwards. I do wish they put more dev effort in support of the CatOS because it will be some time before people fully go to CatIOS. My biggest grip is the port description field is not correctly displayed when a CatOS switch is discovered. It works in CatIOS.

BB

Network_Guru

I am monitoring a national network with 5000 network elements
(see description of a network element in the Orion Admin dox).
Status polling is every 5 minutes, statistic snmp poll 15 min.
MS-SQL Enterprise Database is also running on this server:
Dual P3 930Mhz CPU, Mem,4Gig, 2x18Gb disc, Win2k-Sp3

CPU Utilization varies dramatically from 30-90%.
I would not want to increase polling or add many more network elements
than this. As far as network utilization goes, it is less than %1
of a 100Mbps (the polling load is distributed very evenly over time).
The admin doc has some good info on setting up the polling engine
for your type of network (load calculations)

You need to purchase extra licenses for each polling engine, so you
should be fine running just the one polling engine, with a remote
MS-SQL database. Make sure this server is centrally located in your
network (from a latency perspective). This avoids false alarms
due to high-latency, hi-hop count type connections. Our reporting
improved dramatically when the server was relocated from the West
Coast to Central North America.

Overall this is a very useful product with good support & continuing
development of new features and bug fixes. (I too am waiting for the
CatOS support for polling the port description field, which should
be out in release 7.0).

NG

John

Well I'm not sure if this SNMP problem has been fixed. I just upgraded to 6.4 and added all my switch ports (on one Cat6500) to be monitored and I still get the same UDP buffer overflow errors.%IP-6-UDP_SOCKOVFL:UDP socket overflow from Source IP: x.x.x.x, Destination port: 161. I have been informed that this problem is due to the fact that the snmp queries are all being sent in 1ms and overloading the buffer. Surely Solarwinds can put a delay on this or try and fix this problem some other way? I havent seen other snmp based products behave like this ie Eye of the storm, Openview etc

rmaudsley

Question about monitoring 6509 switch. Where are the two GIG ports in the management module?? Can't seem to find them in the list of interfaces when I bring up the switch.

Any help would be appreciated.

thanks

rix

quote:Originally posted by rmaudsley

Question about monitoring 6509 switch. Where are the two GIG ports in the management module?? Can't seem to find them in the list of interfaces when I bring up the switch.

Our 6509 in Orion shows:
"Port 1/1 - gigabit ethernet without GBIC installed"
"Port 1/2 - gigabit ethernet without GBIC installed"
"Port 15/1 - vlan Router"
"Port 16/1 - vlan Router"

Orion 6.3.29
6509 with dual supervisors, one in standby.
No special config in Orion

Do you have SNMP enable on MSFC?

lionelberton

I was running Orion on an IBM P3 800 Mhz 512 MB of ram and 20 Gb of harddisk. I was monitoring 2000 elements with 5 minutes refresh (I couldn't go below) because the CPU was around 60 % all the time.
Now I am running it on a Dell Bi P4 2.8 C (Hyperthreading) (so it makes virtually a quadri processor) with 2 GB of ram (I would recommend 4 because of the SQL server) and 30s refresh, the CPU load has never been above 10 % !!!!! I tried to use the minimum refresh but apparently the orion engine can't cope with it (although CPU never been above 10 %) and reported all the node has not responding. however none of them ever reported any errors or packet dropped and their cpu value never changed!?!

Lionel.