I was wondering, for the people who work in DC's, what management software do you use to track all your servers, what's on them and what they need to do? Do you tie it in with client billing?
I am in a corporate data center. so all servers are ours.
We use a few pieces of software to monitor the operations of the DC.
Solarwinds to monitor the network
CA Unicenter TNG for SNMP trap management and notification.
I use SysUpTime to monitor and log data for my power strips. I am also responsible for all UPS units out in the trucking terminals...so all critical UPS units are networked and monitored through this software as well so long as it gets an IP and can be polled.
Cernst,
I'd be interested in more details about how you are using the SysUpTime package with your power strips and UPS units -- is it SNMP-based?
Ken
It is completely snmp based. You can setup custom monitors where you specify hostnames and then OIDs. Sometimes the values received are not usable (server tech temperatures come in as 700 for 70 degrees), so you can perform math expressions on them to massage them as you want.
Basically I use it to record data either once a minute or once every 10 minutes. I can then go back and review the data in chart form if I wanted to. It can generate scheduled reports (which unit is going above threshold and how often). It can even forward traps on to other SNMP trap servers...but I haven't really messed with that just yet.
Based on the MIBs for the units, depends on what you can do with it. My APC units have the ability to notify me when batteries go bad...so that is something I watch. I also watch battery capacity, runtime, temperatures, input voltages, output voltages...stuff like that.
A lot of the stuff we use has been written in house for dealing with management, or is already open source and we have made some tweaks to adjust for our own needs.
The big thing that I found lacking was good IP address management, so we've had to build our own system in house from the ground up - it's taken a little while, but with the previous integration of the businesses we've ended up with over half a million addresses now, and tens of thousands of connections which all makes for some interesting management.
I'd agree with the above, SNMP is a great great tool, and pretty much any large scale deployment of equipment I do will be query able with minimal configuration, or scripted to give back results so that we can graph and produce trended data to alert on.
Monitoring systems wise - for traffic we use a combination of netflow tools, and cacti for general traffic graphing. Servers we have a very customised hobbit install with specific tests for all our server types and alerting based on specific criteria.
I love reading this stuff because it gives me a better understanding of all that is involved in having a data center, I know in the long run it will be better for us but some of it is overwhelming.
I have not heard about a single full fledged software but most people use veritas management software for data protection and storage management in their data centers.