Around the office we administrators have a principle we like to call "administering from the Bahamas." The principle...
is essentially that we try to design our systems so that you could feasibly administer them from a hammock on the beach in the Bahamas. Practically speaking, the idea is that when you have servers all over the world you can't easily drop by when something goes wrong. Yet, even if the servers you manage are in the next room, remote management will save you far more than just the effort to get up from the desk.
While there are many solutions out there to aid an administrator in remote management, I've found the lights out management (LOM) built into many modern servers is superior to many of the alternatives. Below I will discuss some of the advantages to lights out management along with some tips on how to get the most out of this type of environment.
Even though each vendor has a different implementation and name for their lights out management (HP's iLO, Sun's ALOM, etc.) at a high level they all offer the same basic level of features. First you get a remote console so you can control the server as though your monitor, keyboard (and in some cases mouse) were hooked directly into it. Second you get the ability to remotely control the power to the server. While these abilities are nothing new, before they were integrated into servers you had to buy one device to give remote console support and another device for remote power. Instead of multiple devices, each with its own IP address and management interface, now you get a single port with a single IP that you can use to manage everything.
Once you go beyond the basics, features vary even within the same vendor. So to keep things simple I will limit myself to some of the features offered by HP's iLO (integrated Lights Out). One nice feature of iLO is the web interface that lets you control power and access a remote console all within a web browser. The web interface also offers diagnostic information on the server's status, along with a full list of the server's asset information--something you can also supplement with extra information from your own organization. An added benefit to the web interface is remote device support. This allows you to map local disks, floppies, USB keys, or even ISO files over the network to the server. This can be particularly useful for booting an install disk or rescue disk remotely.
More lights out management features
Many people seem to prefer the web interface, but iLO also offers the standard terminal-based remote console via an SSH (Secure Shell) interface. This can be particularly useful when you have to fix a machine in a crisis and are away from your normal computer. All you need is a cellphone or Blackberry (which you might carry on you at all times anyway) plus one of the many SSH clients that are available. I can't tell you how many times I have been able to fix a server remotely with my Blackberry when there wasn't a computer or Internet access in sight.
A final hidden benefit to iLO is the ability to more easily manage all of the IP addresses you use for remote management. In the days of remote serial appliances and remote power units, you had to remember which device was hooked up to which server or at least look it up in your documentation, before you could respond to an emergency. With iLO, all remote management for a machine is available from a single IP, so it's easy to keep track. For instance, you could add a new DNS entry called "ilo-servername" for each server's iLO IP. When server44 has an issue, all you have to remember is "ilo-server44."
These are only a few of the benefits iLO can provide. Plus, each vendor has its own special set of features that will let you get even more out of your remote administration. So the next time you are running through the configuration checklist for a server, you should consider adding lights out management. It is one of a few important components that combined help you to administer your machines from anywhere, including (or especially) the Bahamas.
ABOUT THE AUTHOR: Kyle Rankin is a systems administrator in the San Francisco Bay Area and the author of a number of books including Knoppix Hacks and Ubuntu Hacks for O'Reilly Media.