Storage monitoring?

[bigdog71]

Newbie looking for some advice.

In summary I'm looking to be able to develop custom monitoring agents for a variety of storage platforms. I've looked at a few tools (nagios, zenoss, BMC, EMC SRM) etc. We actually have BMC Patrol and portal and EMC SRM and ECC in a few sites. Both BMC and EMC are limited in what they can do when it comes to meeting our requirements. We're at the mercy of the vendor to add the capabilities we need. That usually costs $$$, lots of time or just doesn't happen at all. I work at a medium/large service provider so across all our customers we have almost every platform so an opensource framework is ideal.

My question is around structuring clients in Zabbix. We would have a proxy in each customer's network. We're not monitoring hosts, just storage arrays and SAN switches. Zabbix refers to everything as 'hosts' and I think assumes there's an agent associated with that host. What I need is a 'host' to represent a specific Device (array or switch) on their network that's passively monitored (SNMP or CLI). Can Zabbix be structured this way so a single proxy is collecting for multiple devices and reporting them individually?

Short term we'd only be doing basic alerting on things like failed HDs, fans and other physical components. More advanced things like file system capacities on NAS devices (netapp/EMC etc), replication status/lag, pool capacities would be near term requirements as well. Long long term maybe pulling in some performance data as well.

Any suggestions, comments would be welcome.

Thanks
J

[ingus.vilnis]

Hello and welcome!

Generally you have understood everything correct and everything you wrote in your post can be accomplished with Zabbix.

Yes, every monitored entity in Zabbix is called a host - it does not matter if it is a Windows PC, mail server, router, SAN, website or temperature sensor.

Information from each host can be gathered in multiple ways - by installing Zabbix agent if it is a computer with operating system or you can also use SNMP, IPMI or sensor data to get the values you want.

So in your situation I see everything is pretty simple - you place a proxy in each of your locations and monitor your routers, switches, SANs etc. mainly by SNMP.

Best Regards,
Ingus

[aib]

FYI, BigDog - one of the most important feature of Zabbix is the External Scripts.

You can use any programming language to collect the information which you would like to monitor and save for future usage.

[syndeysider]

Running EMC Clarion (i know right?) and Dell Compellents

I trial'ed 4 different monitoring solutions across a range of "required" devices.

All have cool bells and whistles but non come even so much as close to the extending functionality that is Zabbix Scripts (now also modules).

Essentially any metric value anywhere within your network can be consumed by Zabbix, either via polling or pushing.

I've got SNMP setup nicely for the Dell Compellent, because Dell's nice like that. EMC... not so much. They want to lock you into their proprietary monitoring solutions and their SNMP MIBS are about as outdated out of all the vendors I work with. Maybe it's an age thing (clarion's ancient in terms of storage). What I did manage to get working was a simple polling script that utilized NAVICLI. It does very simple tasks like CPU usage on storage controllers, Disk Status and simple performance (averaged) metrics. The problem is, you need Navisphere Analyzer running which in turn adds a little overheard etc. etc.

The bottom line here, which you probably don't want to hear, is that Zabbix isn't going to solve your EMC problem, well at least that's my opinion.

In terms of "hosts" think of them as logical groupings of simliar metrics. I have some hosts in my setups that have "dummy" ip addresses or agent dns names. These store zabbix_sender values which are pushed in externally from really business applications...

Hit me up PM if you need further help.