Some quick features that I found handy back when I supported HP systems were:

1. Recording of OS panic screens. Basically if Windows blue screened, or Linux kernel panicked, the lights out board would capture it and allow it to be replayed in the browser interface.
2. Firmware logging of events in one central place, including OS, RAID or other events. This was viewable either via tools in the OS, on LCD panels, via the lights out web server, or Insight Manager. Dell appears to have most of this now too.
3. Diagnostic panels. Most of the Proliants had a handy little LED lightboard somewhere that would point out problems, ie a dead fan, bad memory, and so on. I think some of the Dells have this as well now.
4. Raid portability. Being able to just move a hard drive array between servers or even entry level SAN devices came in real handy, both in the lab and for many customers I supported. I had a particular SUSE install on a disk I could toss into nearly every machine in the lab.
5. Out of box setup. With a lights out controller, the only thing that would need to happen in the datacenter would be physical racking of the server and plugging in power and network. With the included tag containing the DDNS name of the lights out system, along with the username and password, the rest of the setup could be completed back at your desk. This includes virtual cd-rom and floppy functions that worked via Java. I believe Dell has mostly caught up here now.
6. Drivers stored in firmware for loading Windows. My memory is a bit foggy here, but there was some feature that allowed drivers to be stored in a firmware disk image, and then used when loading Windows and needing a 3rd party RAID or SCSI driver.

The Insight Manager software package was also decent at providing an overview of the health of all servers in a datacenter, and was free last I looked. I'm not sure what Dell has to match it, though I'd hope they have something by now.