Out of Band Management is probably a term you’ll only ever hear from a server seller, if you’re not familiar with the term, you’ll probably know it as HPE ILO, Oracle ILOM, Cisco IMC, Dell iDRAC, Lenovo/IBM IMM etc. Every server stocked by SMB Servers, (at least at the time of writing) offers Out of Band Management natively.
Out of Band Management, if you’ve never used it before, is a vital component of any server environment and one of the major revelation moments when you first cross over from homebrew servers (repurposed PCs) to true enterprise servers.
If you’ve ever done business system administration, you’ll probably recognise one or more of the brand trademarks above and quickly know what I’m talking about.
What is Out of Band Management?
Out of Band Management (OOB) is a dedicated management system, which, in a server environment, is basically an entire separate operating environment with hooks into many different aspects of your server such as environment sensors, display outputs, firmware controls and a countless other areas as well as using the information gathered to monitor the health, configuration and status of the server.
Through a network interface it then allows you to monitor, control and manage your server remotely without a need for any other physical devices (such as a monitor). The normal scenario is that the OOB starts up when power is supplied, and once it has finished booting the OS can be started.
When running, the OOB, through a shared or dedicated network port, offers a website to the local network which acts as the primary OOB interface. Once logged in (credentials are default or written on the server itself until you change them) you are able to do things like:
- Remotely turn the server on and off
- Update firmware (such as the BIOS/UEFI)
- Remote view/control the display output
- Mount disc images for things like operating system installs
- View server health, sensor information, hardware configuration etc
Fundamentally it allows you to do everything remotely you would normally need to be physically present for on a server, except, of course, for making physical hardware changes.
When would I use it?
Almost never, but when you do need it it will save you hours of frustration.
OOB is an infinitely useful tool when you first get a server, and when something goes wrong.
If you’ve ever built your own servers from consumer components, you know the frustrations of either having to keep a full keyboard, mouse and monitor for every server, or, more likely, physically changing the plugs every time you need to access the POST and boot stages. Secondarily there’s the regular scenario of power cuts or other power events cutting a machine, and not having it automatically switch back on. Or worse, you or someone else accidentally shut it down thinking they were logging off or disconnecting.
OOB solves both of those problems and more, you could buy multiple servers, arrange them far away from your desk because they’re so noisy, plug them all in and never have to physically be in their presence again.
Through OOB, you can check their IP via your DHCP server, access their OOB, fire up the remote control tool for each simultaneously, mount ISOs and prepare and install operating systems for each on your primary desktop computer, while monitoring them simultaneously during each step.
Additionally, if you’re physically away from the servers, such as they’re in the office and you’re overseas on holiday, or you’ve rented rack space to hold them, through OOB you can hop onto their network via a VPN, access the OOB, power them on and off, access the POST environment and take required steps to generally keep them running.
Finally, if you run servers at multiple sites, but don’t have any capable staff on hand to manage them, OOB means you can still maintain most actions as if you had a physical presence, as well as getting prior warning on failing components via the array of sensors and health information.
Where would I find it?
OOB is a standard feature on servers manufactured by the main enterprise OEMs like Dell, HPE, Lenovo, Cisco, Oracle and the like.
There are occasional exceptions, like the Gen10 HPE Microservers don’t have it built in and you have to pay for a PCI-E add-in card. And the Dell Precision range have a rack mount option that are classed as CAD workstations and not servers, but run server gear without OOB.
But if you’re buying Rack or Tower servers running Xeon processors from established business server lines like HPE ProLiant, Dell PowerEdge and others then it’s safe to assume OOB will come natively. There are other non-standard lines that will also have it (like HPE Apollo), for those I would recommend checking their spec sheets, as there’s too many to list and they may or may not natively offer OOB.
Do I really need it?
As is the theme for most of these articles, the answer is, of course, NO. No, you don’t need it. but if you value your time and mental state I would recommend wanting it and ensuring you have it.
If there’s a chance you’ll manage multiple servers, and maximum uptime is important to you, then OOB can be a real life and timesaver in those scenarios.
Occasionally the OOB chipset can irrecoverably fail, the server will often failover to a normal boot and just bypass it and it’s not a big problem. Often by the time this has happened you’ve been running that server for years and never really needed it, and not much will change. For maximum reliability you could look to replace that machine, but OOB is definitely a big want.
OOB is one of the major reasons we started shouting about enterprise servers from the rooftops and subsequently why SMB Servers exists at all. We love it. It has saved us countless hours, headaches and frustration. And it’s one of the reasons we would never go back to homebrew machines.
It’s also not uncommon to need to go through many server reboots in a single session, or to troubleshoot a failing piece of hardware and needing to restart several times. And OOB means you needn’t be physically standing in a noisy server room for several hours waiting for it to POST or fail.
Conclusions
Out of Band Management is a major enterprise feature you’ll come to never want to live without that will allow you to simplify your set up when compared to a homebrew set up with monitors, cables, keyboards, mouse and an array of frustrations.
It’s not a requirement to successfully run a server, but it’s a major time and energy saver that will pay for itself over and over again.
As well as performing the major monitoring duties on hardware, it offers incredibly convenient services like remote power control, remote control, firmware updates and operating system installations.
This is particularly important when you have a power event when you’re physically distant from a server, and is the difference between the server being available in minutes vs the time it takes you to travel to it to simply press a power button. You could be right there on the beach in the image, rebooting and reinstalling your server thanks to OOB.
Most enterprise level servers natively include it under more familiar terms like HPE ILO and Dell iDRAC, and it’s a major feature that sets enterprise servers apart from home brew machines turned into servers.
You don’t need it, but you should absolutely want it.
0 Comments