SAN Monitoring

[djmupster]

What do people use for monitoring / login switch information to help with faults etc.

Our SAN has now grown to about 70 switches 50/50 brocade / cisco. (time we started keeping an eye on it)

[mfrase]

If you have deployed Cisco MDS switches then you must be using Fabric Manager. FM & DM have many logging capabilities and monitoring methods, call home, syslog, accounting, event alarms, traps, full MIB support...

http://www.cisco.com/en/US/partner/netsol/ns514/networking_solutions_package.html

This doc below gives good listing of the wide offerings Cisco has with the MDS's, Items like Performance manager & Traffic Manager do require License, but rest come with base line FM

http://www.cisco.com/en/US/partner/netsol/ns514/networking_solutions_white_paper09186a00800c4661.shtml

As for a system that will give information about both Cisco & non Cisco, standard syslog servers will help some. Cisco MDS does work with SMI-S and there may be management systems you can look at that use SMI-S to monitor/log fabric issues.

[stephen2615]

Considering the money you have already spent, the best bet would be to use Openview Network Node Manager (NNM). Other SNMP managers are cheaper but when it comes to getting the best out of the MDS and Brocade switches, MIB management with NNM is so easy.

SMI-S in my opinion is lagging behind compared to the simplicity of SNMP and I really think SNIA has a provisioning agenda for SMI-S more than a management idea. I use Hitachi's HSSM which is AppIQ renamed. I don't like it at all for switch management. Brocades SMI-S agent is very average and I don't really trust it. Cisco has issues with VSAN's in its version of SMI-S.

There are issues with not using NNM and it relates to Cisco's bizarre time with traps. IBM's Netview does not like Cisco MIBs and it is very tedious to load them.

Syslog is a pain to manage as you have to parse the events and make sense out of them. If you really want syslog information, Cisco switches can send syslog as traps.

So, for ease of use and good information, NNM is the way to go. In a nutshell, all you need to do is load (compile) the MIB's, make sure the information is displayed in the event browser (which is normally not standard) and get your switches to send traps to NNM. I am not sure if NNM can deal with SNMP v3.

The one thing that annoys me a bit is that there are no symbols for the MDS and Brocade switches in the NNM world.

I have spent 13 years working with Openview and I am a Certified Consultant in it but I am not biased towards it. There are alternate products which cost more and don't do SNMP as well. Cisco's MIB's provide excellent information. Brocades.. well, what can I say. It all comes down to your budget. If you want to spend a small amount, give OpManager a go. I tried it and it gives some basic information about FC switches. It does not accept SNMP traps which I believe are paramount to good management.

IF you have such a large SAN, I imagine your IP network would also be big and your company must already have a network management solution. Management of the FC switches is not really different to IP stuff.

Stephen

[inch]

G'day,

For a cheap/free solution you can use nagios and mrtg for performance monitoring.

If you are super keep on the cisco front you can use the PAA2 and ntop! :)