Think installing Mac OS X and a suite of applications, documents, and settings on a single Mac takes all day? Imagine doing it on ten or a hundred or a thousand Macs! Fortunately, tools like Apple Software Restore and NetInstall have allowed administrators to do it a little bit faster, and now Apple has provided an even more amazing tool that gives administrators the ability to roll out hundreds of custom Mac OS X installs in a couple of hours!
Multicast ASR: The Fastest Way to Manage Mac OS X Deployments
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
I’m thinking that things like this, and the upcoming x86 Macs, might make it easier for OS X to “infiltrate” traditional MS-centric organizations. The x86 Macs will be a requirement though, since these companies claim to not want to be tied to one vendor (apparently being locked to a convicted monopolist doesn’t bother them).
– chrish
> The x86 Macs will be a requirement though, since these
> companies claim to not want to be tied to one vendor
> (apparently being locked to a convicted monopolist
> doesn’t bother them).
Why should they switch from a monopolist, who only locks them in with their software, to another, sue-happy vendor, and wannabe monopolist, known for its irresponsibility towards customers, who would lock them in with both overpriced hardware _and_ software?
…and didn’t have a lot of luck. In fact, Apple’s latest patches completely destroy the ability to even create netinstall images. The Apple forums have a fix, but Apple–IN NORMAL STYLE–has completely ignored everyone.
To use the ASR multicast, you need to netboot and run the install from the command line. It doesn’t seem to work as documented. Documentation says it just starts the install from wherever the multicast stream is in progress. In fact, it’ll wait for the current stream to finish before beginning an install. Quite often, at about 80% into the install it will fail out. Because you’re running from the command line, you’ll have to go back and physically reboot any machines you’re multicasting to.
The solution I’m using at this point is to create a basic install image without any security patches applied to do the initial boot/install. Use radmind to create an patched workstation. Use the netinstaller to automatically install the unpatched base, reboot, run the radmind updater, and then reboot again.
From an Admin’s point of view, Tiger has been a _huge_ disappointment. Apple seems to be a lot more worried about pushing out new iPods than getting patches out for Tiger server. AFP performance on Tiger is miserable above 40 clients. I’m going to be investigating other server side solutions for our Apple workstations soon. Apple has fumbled their enterprise support consistently. There’s no way Apple uses the tools they release to their clients to manage the internal Apple network.
Very interesting observations. It will take really wanting to be a mass market vendor to business to fix this sort of thing, and does Apple really want this, with all of its consequences? Not just too busy with iPods, also too busy thinking different.
Wanna bet? They use a variety of tools, but they most certainly use these tools.
Shut up and wait for 10.4.3.
Apple users sure seem to be courtesy to one another when one of their users seem to have a problem.
Its an essential part, its exactly what Windows admins have been doing for years. But its not the only thing.
If you’re an IT manager, you need to know that you can always do new installs. You cannot be totally dependent on one supplier for your hardware. Your software, OS, isn’t a real problem, precisely because of Ghost. Given a disaster, and open source hardware, you can get in a few hundred or thousand machines at a couple of days notice, and get them out onto the desks. You do not have to call up one vendor and see whether they can supply, or are maybe going through a periodic shortage, or are in the course of a product switch, and only have machines which don’t support your version of the OS. Or have just raised prices.
So, if you want to go after this market, yes, easy deployment is essential. So is a second source for hardware. And then, there’s price…
I just use Carbon Copy Cloner and have a deployment knocked out in 30 minutes, using images customized for executives/admin, hr, accounting, creative, editorial… been working perfectly for a LONG time.
Of course when it comes to restoring a users mess, er… I mean ‘data’, that adds the most time. Perhaps when I move this server infrastructure to Tiger from Panther I will try to leverage some of that stuff, but I have other projects to work on… imagine that!
CG
We used the purpose made images thing before, but during the life of an image no one would run updates etc.
Radmind is part of an autoupdating/patching solution.
The writer is confusing broadcast and multicast. Multicast is designed to prevent network flooding. If you are experiencing flooding, it’s because your swithces aren’t running igmp.
I know that some people feel hatred towards Sun but JumpStart was years ahead of its time. You can easily install multiple workstations using JumpStart. No boot floppy needed either. It is true that PXE has been around but the beauty of JumpStart is hard to beat. Even apple could have used JumpStart from openboot if needed.