For 100% uptime, Linux is the wrong choice. That's Solaris with Sun hardware. You can't swap out a mainboard with the server still running with Linux. You can with Solaris. But you'll need to pay highly for well educated people to do it.
I'm not exactly sure what Sun means when they claim the ability to swap out a mainboard with the server still running. But I suspect it it might be a bit of a marketing ploy - and for a few reasons.
1. Mainboard is a loaded term that can mean anything. It can be a complete server in a blade-type server rack. IBM has trademarked the term "blade" so you would need to call your "blade" something else. (Pizza-slice maybe? "Petal" in the new "Advanced Sunflower Farm"? That one even sounds green!
). Anyway, hot-swap backplane systems have been around for years. And Sun didn't invent them. If you pull out a server blade, that server
is down even if your site
is still up on the rest of your rack.
2. Mainboard does not equal Motherboard. A "mainboard" can can contain just RAM modules, or drive controllers. It does not necessarily hold the CPU or CPUs. If you pull the CPUs, your server is down regardless of whether or not you still have AC running through the chassis. If Sun found a way to run a server without a CPU, they would be awarded the Nobel Prize. Or would be if they weren't smart enough to shake down Intel and AMD for multi-mega to bury their discovery first!
3. Hot-swappping components in a live device (note: many companies have systems that support hot RAM, disk drive, and card swapping) does not require a highly trained person to do it. What it does require is somebody that has been shown how to do it a few times, or a technician that can RTFM. There's nothing complicated about it, but certain steps have to be followed in a specific order to avoid making a mess. Just like making coffee: filter first
then add coffee...
Support for high-end devices from boutique vendors doesn't have to be expensive. It just is. I do support for the "big iron." I charge a pretty decent rate. I don't actually need to. But the customer is willing to pay it so I'd be pretty foolish to buck the going rate. And yes, I am "highly trained" but I don't often need to know anywhere near as much as I do to fix a modern server. Most of them diagnose themselves. (i.e. Option Card-01 in Slot-16: hardware failure***
) Some of them even make their own service call! Scary...
Now is using Linux for a web server the "wrong" choice if you want 100% uptime? I think that all depends. Linux comes in a lot of flavors. You have to look at the individual distributions. CentOS, RHEL, and SLED can, and have been used for critical deployments. Their service record is excellent. Using CentOS shows you're done your homework. If you're seriously thinking about running your web farm using Sabayon as your distro, then you've probably been out drinking.
Is BSD better? Depends on who you ask - and more importantly, what expertise is available in your area. If you're in sunny California, look no further than the BSD camp. That's what gets used out there anyway.
If you're in Redhat country, think different.
Getting 100% uptime is more a matter of strategy and redundancy than anything else. All hardware will fail eventually. So the smart thing to do is think in multiples - get two (or more) servers and set them up in a failover configuration.
It's not hard to do. Got a couple of old laptops floating around? Try this:
Setting Up A High-Availability Load Balancer (With Failover and Session Support) With HAProxy/Heartbeat On Fedora 8
Author: Oliver Meyer <o [dot] meyer [at] projektfarm [dot] de>
Last edited 02/21/2008
This document describes how to set up a two-node load balancer in an active/passive configuration with HAProxy and heartbeat on Fedora 8. The load balancer acts between the user and two (or more) Apache web servers that hold the same content. The load balancer passes the requests to the web servers and it also checks their health. If one of them is down, all requests will automatically be redirected to the remaining web server(s). In addition to that, the two load balancer nodes monitor each other using heartbeat. If the master fails, the slave becomes the master - users won't notice any disruption of the service. HAProxy is session-aware - you can use it with any web application that makes use of sessions like forums, shopping carts, etc.
Web link to the full article at:
Like Gypsy Rose Lee used to say. "It ain't whatcha got - it's how you use it that counts,"
(BTW: I agree with you. BSD is a better choice for a server OS. At least for now.