Our 500-server data center is growing quickly and my manager has started investigating using a specialized Data Center Infrastructure Management (DCIM) package to help manage constraints.
DCIM software aims to do everything: capacity planning, asset, config, change management, power & cooling monitoring, network monitoring and alarms. Some of the products she is looking at are dcTrack, DSView & nlyte. They are all proprietary and (IMO overly) expensive. Has anyone had experience with these types of systems?
We currently use Nagios, Cacti & RANCID extensively. We need to improve asset management including tracking cabling - the other DCIM feature my manager is sold on is being able to model power & cooling requirements by dragging servers between racks.
My position is that we are most of the way there with the tools we have and DCIM with all its bells & whistles is not good value, and will force us into inflexible practices that aren't necessarily best for us - not to mention vendor lock-in.
So what other practices, processes and OSS tools can you suggest to help manage a growing small-mid size data center?
I am one of the co-founders for Device42 and we are working on creating a comprehensive DCIM solution, that is not OSS but won't break the bank and there is no user lock-in, we allow easy export of data out.
We don't have power and cooling integration yet for racks, but are working on APIs that would let users add/update data using external scripts/programs. Please check it out and we would appreciate any feedback.
Have you looked at OpenQRM? It doesn't explicitly list cable management, but it seems to have the rest you are looking for.
I agree with you 100% that many of the DCIM tools are very expensive. Disclaimer: I am with a DCIM vendor myself (www.graphicalnetworks.com) which develops netTerrain DCIM.
netTerrain is quote affordable compared to some of the other vendors, but it is all relative of course. We do full cable mgmt, and real time power and temperature monitoring, but you may get a lot of other features that may not interest you (logical mapping, discovery, port status, dashboard, etc)
If you are looking into just a few additional features you may consider developing some of them yourself or getting a point product that is specific to what you are missing right now. For example, you may leverage Intel's DCM module and API for power & temp monitoring and maybe develop something for cable management, if it is data entry you need.
For layer 2, 3 mapping (not necessarily covered by too many DCIM vendors anyways) you can use some open source SNMP libraries that are available.
Hope this helps!
Try adding racktables to your current suite of software to see if that gets you where you need to go. They have a nice demo setup you can play with. While not drag and drop per se, it does help you visualize the racks and Us of rack space to understand how things are physically configured.