Slashdot Mirror
Virtualizing Workstations For Common Hardware?
An anonymous reader writes "We have approximately 20 workstations which all have different hardware specs. Every workstation has two monitors and generally runs either Ubuntu or Windows. I had started using Clonezilla to copy the installs so we could deploy new workstations quickly and easily, when we have hardware failures or the like, but am struggling with Windows requiring new drivers to be installed for all new hardware. Obviously we could be booting into Ubuntu and then load a Windows virtual machine after that, but I'd prefer not to have the added load of a full GUI underneath Windows — we want maximum performance possible. And I don't think the multi-monitor support would work. Is it possible to have a very basic virtual machine beneath to provide hardware consistency whilst still allowing multi-monitor support? Does anyone have any experience with a technique like this?"
Hypervisor?
It's call Norton Ghost.
VMWare View is what you want.
Using Citrix. Not sure if this is what you are looking for or not...
See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
I do. The short answer: Don't.
Just on the interactivity alone, it's slow response, you spend extra seconds loading windows, menus, and after awhile those extra seconds add up to real productivity loss. Virtualization belongs on servers and in labs, where interactivity is less important than raw horsepower. For a workstation, don't virtualize. It's painful.
#fuckbeta #iamslashdot #dicemustdie
It's easy enough to slipstream (lots of) extra drivers and periodically update a master install .iso using tools such as nlite.
"This post is an artistic work of fiction and falsehood. Only a fool would take anything posted here as fact."
Virtualization is not a cure-all (and your approach is wrong, to boot).
What you're looking to do is use the latest, greatest technology for profit(!!!). You're going about it wrong. There are plenty of other, better technologies to accomplish the same basic thing. Proper system imagining/installation via something like an installation server.
When you've got 20 workstations, you're at that cusp of continuing on the path you're on (and hopefully, resorting to a method of consistent repeatability) or deciding on a different approach - thin clients, perhaps. Or maybe virtualization is the right approach - but I can guarantee that there's likely no good reason to virtualize Windows on top of each of the 20 workstations that couldn't be solved with better design.
Honestly, if you're one of multiple IT in a place with only 20 workstations, you're seriously over-staffed. Someone - if not you, someone else - is going to figure this out, and figure out a way to make themselves important and you redundant. Even with moderate consistency and controls, a single competent Administrator should be able to take care of 5 times as many workstations and a handful of servers without too much sweat.
~/ssh slashdot.org ssh: connect to host slashdot.org port 22: too many beers
But only if you have hardware virtualization support.
next year will be the year of the Windows workstation.... 8^)
DRM? No thanks, I'll just get it somewhere else...
NxTop is pretty cool. It is a hypervisor that installs directly onto the client hardware, allowing you to pull and boot pre-configured images over the network. The hypervisor removes the need for specialized drivers and supports dual monitors. It also has the advantage over VMwareView of allowing the OS to sync for offline use if you would like to leave the office with a laptop. Sure VMware has it as an "experimental" feature now, but it is production with these guys. They came and did a demo for us the other day, pretty cool stuff. I think it was affordable too. You can set policies for who gets what images, remotely disable a lost or stolen laptop, etc. Check this out: http://www.virtualcomputer.com/About/press/nxtop-pc-management-launch-massively-scalable-desktop-virtualization-for-mobile-pcs
You're just making it harder than it needs to be. Use Ghost, Acronis, KACE, or any of the other semi-hardware agnostic imaging systems. Failing that, just take individual images of each peice of disparate hardware. Just takes a little one time act for each peice of hardware, and a large disk drive.
Mod point free since 2001
i think ghost/clonezilla is the way to go. you really shouldn't add extra layer's of complexity for no reason.
Do you really think switching to linux will fix your driver problems? The real solution is to use the same hardware across the network.
I mean the having a a cd taped to the side of the case for machine specific drivers might be a little low tech but it prevents confusion.
this was solved a long time ago. Sysprep allows you to bundle whatever drivers you want, and it will just load what it needs on first boot. Combine that with a network imaging solution (back when I worked in that area, we used ZENworks, but there are other options), and ideally network installs of software (i.e. the image should be a base OS and not much else) and you should have limited problems. A new machine type will require a new image, but you can just deploy the old one, add the new drivers, run sysprep and re-create the image. I never had to do mass-imaging of Linux machines, but surely you could take a similar approach for the Ubuntu images?
What is...?
Not always is a common solution the right one. Many times they lack the requisite low level IO needed to do the job right.
Take, for instance, DDC/CI. I don't know what you're doing and that's fine, but in my line of work we have to talk to the monitor. You ain't doin' that on a virtual machine.
Just because it's virtual doesn't mean it's better.
http://driverpacks.net/
I use GSS 2.5 for this reason.
Create SOE in VMWare Server.
Take an image of this via GSS (as the base image).
Use DeployAnywhere to deploy this image onto foreign hardware.
Add platform-dependant drivers.
Take another image of the SOE image in its machine-dependant state.
I've done this for 30 separate hardware platforms (HP, Compaq, Toshiba, Acer etc).
Works brilliantly!
james/logik
It's not cheap so might not be a viable option for a smaller shop, but VMware has been making some very interesting strides in this area.
Check out VMware View, also known as PCoIP (Yes, that is personal computer over internet protocol)
http://www.vmware.com/products/view/
http://www.vmware.com/resources/techresources/10083
Put really simply, each real workstation is loaded with a minimal system and the vmware view clients.
When a user goes to login to a computer on your network, after authentication their virtual workstation pops up (Be it windows or ubuntu) and lets them work.
All of the actual 'workstations' being used are virtual machines, thus are the same unified image you are looking for with one set of drivers.
While I have not tested it with a multi-monitor setup, they claim it is supported.
The one main thing you do lose is full accelerated 3D support, and direct support for old eccentric hardware. (Think ISA card support and non-standard PCI interfaces)
I can say USB support is simply amazing in how well it works.
Clients can even play full interactive flash media and video, and it runs well (As well as one would expect it to work in native OS anyway)
I used unattended on a FreeBSD box at one of my old jobs, since we had like five or so different models of computers. It works sort of like RIS, except it's easier to extend the system since it's all written in Perl and it's all open source. We dumped the contents of an XP disc on the server, then slipstreamed driver packs into the disc directory structure; this catches almost everything but the most obscure hardware out there. Unattended allowed us to run post-install scripts, so we threw in a bunch of other software packages that would install after the OS was done installing, like Office 2007, Adobe suite, etc.
This was substantially better than a disk image; we took care of all of the drivers in one fell swoop, so the only thing we used as a differentiator between computers was how the person used the computer (if it's a student lab computer, we loaded a bunch of stuff like Geometer's Sketchpad, InDesign, etc. If it was a faculty's laptop, we'd load software to operate stuff in the classroom.) We save space on the server, and we save time when it comes to putting together another "image" for a different use case.
But as others said above, I wouldn't virtualize the workstation, even if it eases up on the IT dept. a little bit; just be smart about what deployment method you use. I wouldn't recommend using unattended if you had only about three different models; it's likely substantially easier to just use CloneZilla.
Oh, and use a centralized software deployment system such as WPKG. Your disk images will go stale after a while, in which case you'll have to make sure that you can manage the packages installed on clients somehow.
"Hegelians, who love a synthesis, will probably conclude that he wears a wig." - Bertrand Russell
What you are looking for is called type 1 or bare-metal client hypervisor. Bare-metal client hypervisor's are a fairly new technology with the leading ones "which are still in development" being from Citrix and VMware. They are XenClient and CVP both are expected to be out later this year. Two of the smaller players in this field are Neocleus and Virtual Computer both have a general release product however neither of them have been around long enough to be proven.Hope this helps you might not have a the solution you are looking for today but by next year you should have some good options.
I don't exactly know what you are looking to accomplish, but aside from spending money to make the machine identical, you can look into LTSP or the OpenSuse version Life. These allow your normal workstations to boot over the network and then depending on what you are looking to accomplish you can have them call a Terminal services session or just use the Linux distribution that is loaded. 20 Workstations are a breeze for 1 4CPU/8GB ram server especially with the progress of local apps on the client side. Have a look :)
Existing deployment tools from Microsoft already do this. You need the WAIK, which is a free download from Microsoft.
You need to create a generalized image. If you get all the required drivers for all your hardware into the driver store, the drivers will be found during install. You can also deploy from PXE boot using WDS with a generalized image...
There are a few caveats around a few drivers that aren't designed properly for Sysprep, and applications that aren't designed with sysprep in mind, but otherwise it's quite slick. You can script the installation of these exceptions to occur later on during deployment using unattend.xml and RunSynchronous commands though. You can also supply your licence key in the unattend.xml file.
About 90% of all Windows deployments are sysprepped by OEMs or by corporate IT folks....
Please read the documentation, the tools are quite flexible.
I just did this to use a single image on my company's multiple versions of their standard hardware (they use everything until it dies a horrible death). I used to use nLite to automate the install and just slipstreamed the drivers in but a driver for a new model's raid controller would not integrate so we switched to Acronis and it worked the first time. Imaging from the Acronis-prepped DVD now takes 15 minutes which used to use to take 45 minutes when we installed with nLite. Most of the savings came from not installing all the apps we use. I had nLite automatically installing a lot of stuff like, .net, hardware support apps, and A/V with multiple reboots until it was all completed. Now all the apps are installed on the image stored on the dvd. I know nLite is free but we were willing to pay for the cut in deployment time and administration.
For the Windows side (I'm assuming XP), put DriverPacks (http://driverpacks.net/driverpacks/latest) in a folder on your disk image and use SPDrvScn (http://www.vernalex.com/tools/spdrvscn/) to add them all to the driver path. You'll need to use SysPrep and link to the each mass storage driver in the SysprepMassStorage section of sysprep.ini to support RAID and AHCI storage controllers (otherwise you'll need to use IDE emulation in BIOS for a performance hit).
Linux should be pretty resilient to multiple hardware configurations on its own, but I guess that depends on the distribution.
Your time/money is probably best spent standardizing your desktop hardware at this point.
Wow, such terrible advice from slashdot. The easiest way to move Windows OS from one machine to another when their are hardware differences is to get your self a copy of shadowprotect and use the HIR (hardware independent restore) option. Google it. Virtualising is not the best way to by a long shot to do what you are trying to do.
Really?
Can I run a Windows 7 virtual image (Virtual Clonedrive) on an Ubuntu PC somehow? On a P4/2.6Ghz/1GB-RAM machine? Fast enough to run Visual Studio 2010 and test Silverlight apps? How?
--
make install -not war
VMWare Workstation 7 has support for multiple monitors.
We've been looking at several options for this type of thing at the University where I work. We've got one lab currently running LANDesk, and another that is running off of VMWare. They each have their own advantages and drawbacks, you've just got to decide which is best for your situation.
In the multiple computer labs that I maintain virtualization is out of the question due to the performance hit being too high considering the usage for the two of the labs (Photoshop, streaming video, audio decoding, some really heavy javascript stuff). The rest of my labs lack the funding/justification to setup a VMWare or LANDesk backend and they get the hand-me-downs when I get new hardware for the larger labs. What I've done instead is I use Altiris (now Symantec) Deployment Solution. I've put together a basic lab image (MS Office, Firefox, anti-virus, Windows updates, etc.) in VMWare Fusion on my Mac. I then deploy that image out to the rest of my labs as they only need Office and internet access. That way I only really have to maintain two lab images. One for the two Photoshop labs, and one for the rest of my labs. When important updates come out I update the VM, and have my deployment server push the updated image out to all of my machines in the middle of the night so I don't get in the way of the students being able to do homework and other assignments. The nice thing about Deployment Solution is it has an option for hardware independent imaging where it removes the existing hardware abstraction layer (HAL) and injects drivers for whatever hardware the image has been deployed to. You do have to maintain your driver database and make sure that you get updated drivers for new hardware, but this has worked flawlessly for me for quite a while now and I'm imaging against...I think five different sets of hardware (they mostly differ in the motherboards, no video cards in the lab machines beyond the integrated video) with the one base image.
This space for rent...
Am I right?
Cut to the chase.
You have Client machines - not all are going to be the latest or the greatest in hypervisor tech., [you do what you have to do to keep things afloat]. Consider, Thin Clients from a myriad of hardware offerings, less headaches and better Server hardware will keep you way ahead of the curve and lessen your - footprint, exposure and budget.
The caveat is only if your Clients run AutoCAD, heavy graphic intensive programs or major databases, programming.
Windows, UNIX or Linux - or all, "pick your poison" the rest is academic.
Good luck.
~hylas
Yes you can do it with VMware ESXi, if and only if the hardware supports it.
Excuse me, but please get off my Pennisetum Clandestinum, eh!
A small or entry-level mainframe could do it, consolidate every single one of the boxes into one. That or find something with say 32 cores.
C|N>K
Its not worth the security risks as it will be impossible to detect and delete viruses and malware that hide with in it unless you disable it in the bios and do a format c:.
You have run into the problem that many companies face which love to lock things down. Slashdotters hate this but its nice to have everything the same hardware and software for reasons like this.
Someone mentioned Citrix clients and vmware or virtual box players but they really really suck and consume incredible resources on desktops. Try to standardize on common desktops and create a script to use Windows update for the latest drivers or keep them on a USB flashdrive. A pain yes, but this is what your customers want. Anything else makes the setup look bad which ultimately makes your boss look bad which makes you look bad if you have a buggy slow activity and responsiveness from Citrix clients or virtual machines. Consumers judge responsiveness in quality more than performance. If it takes a few seconds for a menu to pop up they will think they are on 486s.
If the hardware is nearly identical you can setup a profile with Windows Update for drivers disabled.
http://saveie6.com/
Dump these motherfucking windows boxes.
How this fucking OS annoys me...
VMWare makes a product called "VMWare View". It's basically a thin client that connects you to a VM running on a server. Most Windows thin client environments boot a RDP thin client that connects to a Windows TS, but this approach gives every workstation their own Windows environment to screw up. While not exactly what you're looking for, it will provide a driver agnostic approach to running a workstation. (although I would go with the nlite idea proposed above)
...since you're apparently a capable Unix admin, ntfsclone.
But that doesn't apply here. That applies to viruses screwing up your Windows, but when hardware dies, particularly the motherboard, that image isn't going to help much.
Don't thank God, thank a doctor!
Wow, that is so insightful. Do you have any more advice coming from that closed mind of yours?
Well you probably should do a couple of things, possibly more, just to be safe / conenient for varying possible future use scenarios.
1) Make an image copy of the entire drive, and any others that are referenced by your configurations. Boot sector, partition table, C partition, other partitions / drives, the whole set. There is really no general substitute for having a copy of every single factor that could affect your ability to recreate the system exactly as it is if you need to do that on a physical machine or with some future set of virtualization tools. Since it is next to impossible to rebase / reconfigure applications that have configurations referring to paths under D:, E:, F:, DVD/CD drive as O:, whatever, you'll want to note all the mappings that could be relevant to making the applications work again and copy that data too.
2) Look at the free "disk2vhd" tool from Microsoft's sysinternals site. Maybe it can help convert your physical C partition into a VHD image which you could potentially eventually boot with something like "Windows XP mode" or Microsoft Virtual PC or Hyper V. Read up on some physical to virtual scenarios using Virtual PC and Hyper V and XP Mode and see what is most likely to work for you. There are various good technet / microsoft / msdn / 3rd party FAQs and blog posts about the good and bad points of doing physical to virtual mappings like that with their various tools.
3) It is possible you could make some use of the Windows AIK or MDOP tools to help your physical to virtual conversions. One thing that is commonly done before capturing an image from a physical machine before virtualizing it is "sysprep /generalize" which takes out some of the machine specific device drivers, licensing activation data, etc. so that the resultant image is more generically transportable to a different machine or VM. YMMV. The blogs / recipes online above will guide you as to the best options.
4) Check out Virtualbox the free VM system from Sun/Oracle. Read their forums about some physical to virtual capturing scenarios. I'm often more impressed with the functionality of virtualbox than microsoft's virtual pc / XP mode, so maybe it would be a better choice for you. Though the tools to do p2v conversions are kind of weak in both camps, nothing truly a click once automatic process.
5) There are probably some good ways to do physical to virtual conversions with a LINUX OS too; the qemu/kvm hypervisor is pretty effective at virtualizing XP in recent versions of LINUX like Fedora 13 beta or Ubuntu 10.04 beta 2 both of which are newly available, though the qemu/kvm virtualization has been working well for years. OpenSuse11.2 should work too. Anyway there are various tools you can use to capture the images of the XP C partitions and other partitions into QCOW or other such formats that can be used with the VM software to run the virtualized system. Again device drivers loaded into the physical XP system will often possibly be problematic so either remove them manually or sysprep /generalize the physical OS or just try booting the VM into safe mode and then getting rid of the old drivers. Whatever works.
6) Of course the easiest solution probably hasn't been invented yet, and next year's VM systems might not even be compatible with some of the disk formats and configurations todays VM systems use, so, again, that's why it's good to keep a full image or physical copy of the original drives handy in case you want to convert them again later.
As others have mentioned, you don't want to go down the rabbit hole of virtualization just to manage 20 computers in an office.
Altiris products are worth considering. The Client Management Suite is pretty terrific for managing lots of dissimilar clients. The rebranded "Backup Exec System Recovery Solution," doesn't do as much, but also works fine with different hardware clients. I haven't bought anything from them since Symantec bought them, but we loved Aliris before that.
If that's too rich for your blood, a SOHO WHS box may be enough to cover your windows machines. Ubuntu machines are easy, you could go as barebones as something like crontab, dd, gzip & rsync on an NFS share. Or you could do something fancier.
sysprep is not a 100% thing and some drivers have there own control planes / back round apps that may or may not be loaded right after sysprep.
This is an excellent example of not identifying the problem correctly. You have 20 workstations, all different, but you need them to be common. Rather than investing in a software solution and all of the complications it will bring, just fix the problem. Lease 20 identical workstations for 3-4 years. Build one image.
Stop looking for a stove+fridge combination, buy a fridge and a stove.
Seriously, you will spend less money and have a faster result by buying two machines, if you need both environments - unless you're talking about many hundreds (perhaps thousands) of machines, it's difficult to justify building a merged Windows / Ubuntu SOE in terms of delivery architecture. What would the merged SOE look like in terms of budget, after it's filtered through a bunch of consultants? Ubuntu doesn't take much in the way of hardware to drive, so you can get the lower spec gear to get decent performance. YMMV of course, but I'm willing to bet you'd get a cleaner result by keeping the two environments separate, and save a lot of money by avoiding the merge altogether.
Do not mock my vision of impractical footwear
One answer is a terminal server. There are a couple of drawbacks and its not a solution for every office, but the advantages are many:
- Workstation drivers/quirks are far less trouble or made moot altogether
- You can set security policies, do application installs, and just generally manage things from one place
- Backing up everyone's data is way easier
- Users can login and gain access to their files from any workstation (even allow VPN in)
- A dead workstation can simply be swapped out with another with no hours of restoring/customizing afterward
You mention that both Windows and Ubuntu are in use. If standardizing on one or the other is impossible, then have your terminal server run one of them virtualized. Virtualizing Windows is probably the better choice since then you won't have to bother with licensing/driver issues when you upgrade the server. There's very little virtualization overhead these days with modern hardware running KVM or Xen.
The drawbacks might be:
- Any graphics-intensive applications are not going to perform well (video playback is usually fine, though)
- Access to local storage (usb keys, cd-rom, etc) *can* be problematic
- Dunno if it's possible to run Windows dual-head while its virtualized *and* being run on a terminal server
- The terminal server needs to be robust against hardware failure (RAID, redundant power supplies, etc) to lessen its impact as a single potential point of failure for the whole network
Your last line was absolutely spot on.
One image per machine is the wrong way to go about dealing with with hardware/driver conflicts. The effort and money could be better spent on gradually replacing odd-lot hardware with something standardized. For the windows machines, I find that 10-30 RDP sessions on one VM server, with 4-6 servers per virtualized host, now that works. I've had around 300-600 sessions virtualized across perhaps 10-20 servers in a dozen heterogeneous sites over the past 5 years. No catastrophic losses, not one customer has complained about speed issues or canceled the service for any reason, and a grand total of ~20 days of downtime across all servers, primarily to reboot for patches and updates. The kicker is that these would be Windows 2003 + 2008 servers on Windows Server 2000, 2003 and 2008 hosts, plus a handful of XP single-user images for special circumstances. Egad, really?!?! Windows VM on Windows hosts? Yes, really. Only very recently have I switched some hosts over to Linux or ESXi, and frankly, there's very little meaningful performance difference; the switch was undertaken largely to free up license slots on back end servers which have no user access. I love Linux and open source but the FUD and misinformation floating around regarding windows reliability and performance is .. well it doesn't square with my experience. But, I'm tired of arguing about it with people. Do some of your own benchmarks and see for yourself.
Even inexpensive server hardware (4k-8k range) is overkill, performance-wise, for a well-balanced, optimized VM internal cluster with sensible backup and monitoring controls. This arrangement has allowed me to more or less single-handedly manage the entire user base referred to above, for the entire time. I've reached the point where I cannot handle the new business that is steered my way by present clients via word of mouth. I have evicted 6 or 7 In-house "IT guys" who were sitting on their rears playing WoW 4 days a week, tending a tiny garden of 15-30 machines.
I use Kubuntu for my workstations. I load them from a PXE boot server, which installs using a combo of a kickstart file and preseeding. The kickstart/preseed config looks to a local mirror to install from. That local mirror is run by apt-cacher-ng so it's always up to date. If you're trying to get maximum performance, don't bother trying to use a VM for Windows. I do that as well, but it's only to run legacy apps we don't have Linux versions of yet. I've gotten to around 95% Linux only use, so it's not a big deal for me. The Windows world has tools to help you with that problem, notably sysprep. There are others I'm sure, but that's not my area of expertise. Let me know if you want more details.
Why is it so hot? Where am I going? What am I doing in this handbasket?
Sounds like your problem is in having to support different drivers for every different piece of hardware.
How the heck is Virtualisation going to help you? Even if you have a Virtual machine that emulates a standard config of hardware what are you going to run the Virtualisation software on? You are still going to have different hardware to account for at some level. Whether it's a hypervisor or the OS itself somewhere down the chain the differences in hardware are going to have to be accounted for.
I don't get what you are trying to accomplish.
OPSI allows windows to build itself across the network via PXE, and allows deploying of apps also. It pitches itself as an option for non-identical hardware where cloning works poorly or fails. It's also OSS.
As an aside I'm attempting to combine OPSI /w FAI and GOsa (an LDAP management platform) to manage workstations, servers, services (such as Samba, DNS,DHCP, ftp, Asterisk, Groupware (Kolab, Horde, SOGo, phpGroupWare... you can make it manage just about anything), Nagios, Netatalk (for Mac file/print), Squid, etc.. etc.. etc... Warning... this way leads to madness. It's insanely difficult, and I'm in the process of trying to document and/or automate a lot of it to make things easier. See https://oss.gonicus.de/repositories/gosa-contrib/squeeze-install-scripts/trunk/ for my efforts so far.
At least on the hardware, I see two potential hardware related issues. One is i586 vs. x86_64. You will have to make an image of each type. The otther problem will be your IDE/SATA Controller. The image will have all the IDE and SATA drivers you need, however, you may have to use PXE Boot to cause it to rebuild the initrd for your IDE/SATA Controller of choice. The OS should take over after that.
If your package manager is worth its salt, you should be able to load a list of uniform install packages from a text file. Config files can be rsynced. Keep a Kerberos server handy, it will save butt loads of time.
Desktop virtualisation is hypeword of the month. Don't get suckered into it until you understand the whole concept.
It's financially feasible only after you have 200+ desktops which you turn into virtual machines. With less, it'll just cost more.
Your case might be slightly different though.. you want to virtualise the OS on same hardware the user is currently using.
Remember that hypervisor adds overhead, you lose performance always.
It also creates some funky clock skew issues sometimes, and your virtual machines might have hard time updating their virus definitions.
For 20 machines, I'd say its more trouble than worth, unless you need to do it for compatibility reasons (like running older autocad on windows 7 64bit).
There are no atheists when recovering from tape backup.
Well, then it's really tough to put you into a right Dante's Hell circle. It's either #1 or #5. :)
Honestly, if you're one of multiple IT in a place with only 20 workstations, you're seriously over-staffed.
"Honestly", you're making a lot of assumptions and have invented a scenario where a)the story writer is IT AND b)The company is only 20 people AND c)They are overstaffed IT-wise. Do Slashdot posters ever listen to how stupid they sound?
Maybe he's a developer or similar user at a small/startup company where they are the most technical people already. I was hired on to my first job because the engineers were tired of playing tech support for the rest of the company. Or, maybe they are a workgroup in a company of hundreds or thousands, and the primary IT group is incompetent. Etc.
Please help metamoderate.
I call bullshit on the troll moderation of parent (also my post).
This guy has a very niche problem that:
(1) should not be solved with virtualization,
(2) is only because of some arbitrary and tedious restriction on how Windows installs drivers, and
(3) does not belong on Ask Slashdot.
The latest versions of Ghost are the same as Backup Exec System Recovery, except that BESR allows you to back up/clone servers.
BESR is really good software. I use it quite a bit; I back up all my home servers with it. I've used it to perform P2V's of servers when the normal Platespin/VMware Convertor doesn't work for whatever reason.
BESR/Ghost allows you to take a full snapshot of a disk or a full machine. It's very fast. Restoring is very easy; insert the bootable Vista-based CD, and restore from a local disk or network. You can load drivers from USB sticks or CD's, or make custom bootable discs with your own drivers.
You can take snapshots, and create incremental snapshots on top of those. The incrementals are super fast. You can mount any snapshot as a drive letter, if you want to.
It has a "restore-anywhere" feature, which will put Windows into a sort-of "OEM" mode, so when it boots up for the first time it will run the hardware detection routine. I've been able to move Windows to different hardware real easy this way. You can also use it to move from a Virtual machine to a physical one.
I can't really say enough good things about BESR. It just works, it's fast, reliable, and restorations are super easy.
You can download a trial that works for 90 days from Symantec's site.
- It's not the Macs I hate. It's Digg users. -
First of all, the easiest solution: buy consistent hardware. Get a bid from your systems vendor for a workstation build and a laptop build, and buy them in lots of 10 or so. (The bigger your order, the better rates you can get, natch.) You can either stick with the OS (and drivers) the vendor provides and just install your own crap on top, or make a Windows image with all of the necessary drivers slip-streamed in. This is how most companies do it-- it's easier on you, and it's cheaper on the company.
If that's not possible, just make a Windows image with all of the drivers for all of your hardware slip-streamed in... extra drivers can't hurt, they just use up a bit of disk space nobody will miss.
Note, however, you really should just switch to Windows 7, as it has built-in drivers for 99.9% of what you'll encounter day-to-day, unless you're running some very weird custom hardware. I'm guessing that you're still using an old XP SP2 or SP3 disk, otherwise you wouldn't be running into this issue in the first place.
Of course, it also sounds like you guys aren't really that big an operation. You could just keep a USB key handy with all of your drivers on it, and reinstall boxes as they come in/are needed.
Comment of the year
In fact, I use a hard drive that I pulled from a dying laptop as my 'traveling Linux box'. Any machine (desktop or laptop) which can boot from a USB drive is a happy host for Linux with no, or little, change to the system (sometimes resetting the X display for odd video cards). The biggest 'problem' is that, under Ubuntu, every ethernet/wifi card gets a unique number.... Right now I'm up around eth10 for the newest box.
(there are relatively simple ways to fix the ethN numbering oddity, but I'm just too lazy to write the scripts).
Free Software: Like love, it grows best when given away.
It was only for personal use and as experiment. I set up a small Linux partition on my hard disk. Then, I did a minimal Linux install, with just the bare minimum + X server and VMWare GSX Server. I set up X to load the VMWare client program full screen on start. Then, I created a virtual machine (Linux in this case) using the un-used partition on my hard drive for drive space. Then, I configured VMWare to start the virtual machine at system start. So, upon logging in and starting X on the host machine, I'd get a VMWare client ready to connect to the booting VM.
It was a clunky solution for a dumb problem, but it worked, and performance wasn't as bad as you'd expect.
Commonality of hardware is what your going for, and to try
and use software to make up for it results in more work for you/the admin.
Having a base hardware profile for your "workstations" is important , and if you can have hardware commonality between your workstations, and servers, all the better. Your generic pc's can be whatever...
If your pushing your work stations hard, your on a 2 year (or less) upgrade cycle, so unless you overplan, and building expandability, your gonna be spending an average of 2-5k per upgrade cycle, per machine.
As a straw man (since these are my keystrokes), I would pick a server/ws vendor who had the fewest common SKUs for their machines. My fav so far is introtrend/uniwide. They have this clever little daughterboard that sits ontop of the main board that allows an additional 2 CPU sockets, and an additional 16 dimms to be added. You can run the 4 CPU boxes in a 2 cpu mode, with a little bit of fiddling. The advantage is that you can get a bunch of 2x core 8000 opterons cheep of of ebay, and then as the users wine for more power (a year down the road), swap them out for 4x core, or double them up, placeing 4x CPU in each box. Because this is running one generation behind, your gonna save some coin. You get 2 16x slots, that can fit double wide vid cards, not to shabby.
You stock the same spares for your work station, as your servers. If you have to run win7 , then you can only use 2 physical CPU's, (but all 32 memory slots), or of you install 2007r2 you can hit all 4.
Vid cards are another matter, if you have "performance issues" , then your gonna have to get set to play musical drivers, its a fact of life. You can limit the family and supported chipsets (nvidia) for the ones you buy, but only if you get the backing of the people who hold the purse strings.
If you explain to $ holders, that you can save the cost of 3 or 4 computer purchases over the equipment lifetime, and that the upgrades will be "low cost", just getting additional memory and CPU, vs a whole new machine.
You can also let them know that you can keep one spare motherboard around, and fix your servers, or a workstation.
I happen to have 5 of these boxes (the 940 sockets) at home, and should be setting up a few more as they get retired from datacenters, Photoshop rocks with 60gb of memory.
In one place that followed this plan, these machines are in front line use, after 5 years. They started off as 2x cpu's, went to 4x cpu's, the vid cards got SLIsed, than put into slower boxes as better vid cards came out, and then into the servers.
You can do this with any high end bit of kit, just set standards, and follow them, high end server and workstation boards tend to have a multiyear life, the mid end commodity pc's tend to last a year before they fade away . The Tyan n6650W was out in 2008, and should be around for a few more years.
http://www.overclock.net/amd-build-logs/357378-dual-quad-opty-sli-build-tyan.html
If you are unable to set standards and follow them, your kinda screwed anyway.
Debian and KVM work great and no reason it shouldn't scale, KVM has some great network admin tools. http://pve.proxmox.com/wiki/Bare-metal_ISO_Installer
If taxation is legalized theft, then Capitalism is a prolonged rape followed by a slow death.
1) best option is to lock keep hardware models limited, and only change models when absolutely necessary. i.e. For example you may have some categories of devices you require -High performance laptop -Ultra Portable Laptop -High performance workstation -Standard Desktop Then assign a specific model to each category, and only purchase that model. 2) go to Windows 7, get driver updates from WSUS or internet. this will pick up large majority of drivers
This sounds very suspicious to me. It states "hardware failures or the like." He has 20 workstations and suffers from constant hardware failures? And what does "or the like" mean? Why are you re-imaging your 20 computers often enough to warrant this problem? I have a shitload more systems than 20 (>300) and I spend way less than 1 hour a week installing windows drivers even though I experience plenty of hardware failures and I install windows from scratch with an sif file. Also, multiple monitor support in Windows is so simple the end user can set it up. Something is very fishy with this post.
How about creating an SOE (standard operating environment) DVD with driverpacks from driverpacks.net? To see an example of this, do a google search for my little baby Winborg XP. Anytime we get a new machine that will host Windows, we just use the SOE on it and the drivers question is not an issue. We also install our essential 3rd-party apps as a part of setup and so it really is a dream to use. Stick the DVD in and come back in an hour with the PC totally ready to go. Takes 2 minutes to individualize it for the user it is for and wallah! The best thing about my Winborg XP is that the latest edition is a builder edition and so u could just modify that for your needs, but this is the lazy option, i fully recommend getting into unattended installs, beats the Ghost method hands down due to the drivers problem being solved. Each month i make a new SOE that incorporates the new hotfixes from MS, any updates to the 60 or so addons we integrate into the SOE, and any updates to the 3rd party apps. I have even scripted the SOE to be built automatically, so its a pain and trouble-free process. Having said all this, if i didn't know this could be done, i would go with a VM method for sure :)
Reading the original post, the first question that came to mind was: why is the hardware configuration not being controlled!?
In another life managing labs, configuration control of the hardware was step #1 towards sane management of any group of computer systems. In the current life of managing thousands of servers, this remains the case, regardless of the OS.
The ONLY reason why folks have a hodge podge of hardware is because someone is trying to be cheap and "save money", nevermind the countless hours spent addressing issues that arose due to bad decisions and/or lack of planning at the start.
Suggestion #1: Review your current hardware/driver issues. Remove or replace components that are having hardware issues. Your lab shouldn't be changing hardware left and right on you. If it is, that's another problem. Once you've moved all of the systems to a nice overlap of compatibility, restrict all further future purchases to hardware that is well supported by all OS(s) in your employ.
The second issue that I'm seeing, which needs more information: what is your intended support list of applications?
Most workstation labs have a list of supported apps. Anything not on that list is "you're on your own". You can't please everyone. That was the case when we didn't have literally hundreds of thousands of potential component matchups. It most certainly isn't the case now. Obviously, if you can reduce your list of operating systems down to just one, you will be in a happier place.
Now, regarding the issue of destkop virtualization, as you've described: you are SOL. ALL desktop virtualization effectively requires the host OS to be running a GUI and the guest OS to be routing their "display" through the virtualization technology through to the host OS's GUI.
If you have to go with virtualization: VirtualBox. You can use either Windows or Linux as the Host OS. Supports both as the Guest OS. It is free. Runs fairly well. Allows reasonably good USB passthrough.
If you don't mind rebooting to change OS(s), another option you can have is to image all the workstations as dual-boot workstations. People can just reboot into the OS of their choice. You get max performance. Regarding drivers, I'm not sure I see the problem. Keep a master image updated on your central server, with all drivers. Each night, auto-reimage your boxes, so they start the day out fresh. I mean... this is a computer lab. Re-imaging each box EVERY night is just good common sense.
Final solution I would offer: Net boot your workstations from a central server. This has the advantage of having each user's intiial experience be on a clean/safe system of their OS choice. Since the image being run is from a centralized server, you only need to keep drivers up to date on one host(master).
Anyways, there are many ways to solve the problems you've listed. The best is to attack the root of your problems; standardize the hardware configuration and prevent future configuration drift.
Winged Power Photography
what the F are you running that is forcing you to keep Windows around then?
Donald 'Duck' Dunn: We had a band powerful enough to turn goat piss into gasoline.
Thin Clients might be a good solution to your problem, unless you are doing really graphics heavy stuff (image manipulation/video cutting). For computational work, software development or common business tasks the Sun Rays are great.
And the really great part is that sun virtualbox can read the vmware virtual machine created by that tool.
and vmconverter can eat a lot of diskimages format also (BUT NOT ALWAY THE VERY LATEST, check before spending time on it!!)
Before you start with any tool it is nice to clean of any unwanted software and restore points, clean the trashcan (Crapcleaner tool), and try to fill all unused space on the disk with zero's.
BTW, I learned the hard way that truecrypt is incompatibele with any on the fly diskimagers.
Have you tried the Microsoft provided Solution Accelerators? free, scenario-based guides and automations designed to help IT Professionals who are planning, deploying, and operating IT systems using Microsoft products and technologies.
This one for deploying images/apps/user migration with hardware driver management - Microsoft Deployment Toolkit (MDT) 2010, source code is available too
Why root around with hypervisors, etc when all they're going to do is introduce performance problems, an additional host OS to keep updates, etc.
The new MS deployment tools (MDT 2010) are good - you can split the drivers and core OS out so that you don't need to keep creating new images for new hardware - just upload the upated drivers to your server, and bingo...
I run: Windows, OS X, Linux, FreeBSD. Just because you have a hammer, doesn't mean everything is a nail.
You should really integrate all of the drivers into your Windows installation. Windows will install the drivers the same as the drivers that come with the OS. It's not too hard and it's probably better than using VM's.
Basically you can create a directory tree in your existing installation and add the path to the "DevicePath" string value under "HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion". Windows will expand it's search for matching hardware INF files to include the extra directories. There are two good articles on doing it: MSDN Article,MSKB Article.
Another method I have used (if you know how to edit INF files) is to manually edit the INF's to use a specific cab file as the copy source (under the SourceDiskNames section). You can also combine and streamline them also. Then just copy the edited INF files into "%WinDir%\INF", and the driver files into the CAB and Windows will use them.
Once you are done just make sure you have Windows update it's INF indexes, which will speed up the driver installation for your driver packages. Open Add Hardware from the control panel, tell it to install manually from a list, and then select "Show All Devices". It will take a while to build the indexes, and then you can just cancel the wizard.
The clash of honour calls, to stand when others fall.
Unattended isn't so much use if you want to put anything other than 2000/XP on your box.
>> But try having a thousand, or ten thousand, trying to access the same terminal server
who the hell would ever do this? this is a non-starter scenario, like arguing against commercial airliners as transport because you could never fly safely or even get airborne with a thousand or ten thousand people inside. With 1,000 or 10,000 users you split the pop up into multiple servers, not try to cram them all into one machine. Ridiculous statement.
"BTW, I learned the hard way that Truecrypt is incompatible with any on-the-fly disk imagers." [Edited to be more readable.]
I don't understand that. I've never seen any incompatibility. Could you explain how Truecrypt is incompatible, and with what disk imagers?
This guy is a troll
Para-virtualization is what you need. The virutal monitor runs on the bare hardware and virtual machines (Linux, Windows) can run in parallel on top of it.
...and choose hardware from a vendor that understands this - most vendors will sell a "pro" line that is substantially the same as a home desktop version, but they'll guarantee production of that model with that hardware for x years - e.g. Dell Inspiron vs Lattitude models.
DOn't get me started with manufacturers who think it's a brilliant idea to swap 3Com for Broadlink NICs mid batch...
Title: Virtualizing Workstations For Common Hardware?
First sentence: We have approximately 20 workstations which all have different hardware specs.
Bad form to answer a question with another question, but WTF?
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
Just upgrade the machines from XP to 7. 7 not only includes sysprep (the tool to give you that initial setup wizard to make each PC unique as far as SIDs etc) on every machine in c:\windows\system32\sysprep but they have re-written the Hardware Abstraction Layer (HAL) since XP so that it is really good at re-detecting hardware differences on boot. You couple that with the fact that it also now installs every driver from the install DVD on the PC (along with all other install options) it means that you can do the following:
1.) Get a machine just the way you want it
2.) Run sysprep with the generalise option
3.) Clonezilla it
4.) Install that image on any piece of equipment forever more
I have taken images this way on desktops and used them on laptops and vice versa. It seriously just has the drivers for 90% of the hardware involved and I install a driver or two here and there when I need to after the rebuild when it doesn't.
So the answer isn't go to Linux or virtualise things - it is just update to the current version of Windows.
The newest version of remote desktop supports multiple monitors, can't say about Ubuntu, but considering all you need to run remote desktop is... well, networking and video output, you can't really go far wrong. Start it up with a commandline argument to hide the bar at the top of the screen.
Not sure what your budget is, but you might look into Pano zero clients as a solution for general purpose business workstations. Basically you put up an esxi / vcenter server, and P to V your workstations from the hardware to the esxi server using VMWare Converter. Very easy process.
Then you deploy Pano "zero clients", which are little black boxes with no moving parts. You then fire up the Pano server VM. Pano distributes their server software in VM format. You simply use VMWare converter to add it to your esxi box. Then you run a simple web based setup on the Pano server. Finally you go into the Pano server and associate the Pano device with an AD account. Walla. That user logs into the Pano device assigned to them, and they get the correct workstation. If they want to go work at another desk, just assign them a different Pano device. Instant roaming profiles without all the hassle.
Pano also offers a dual monitor attachment. I have not tested this, so I can't say how well it works, but there is one available.
Some other benefits are that if the building power blips (assuming you do not have a backup generator for your entire building), the Pano devices connect right back up to esxi, and no work is lost (assuming you have adequate backup power in your server room). DR is also fairly easy. You just use VCDR or VCB (being phased out) to restore your workstations.
There is some minor power savings as well. The Pano uses only 2.5 watts, whild a Dell Optiplex GX520 uses 250 Watts. You have to factor in the power use of the server as well, so in some cases you might actually use more power in the zero client scenario, depending on how many VM Workstations you can cram onto your esxi boxes.
I worked on one of these Pano boxes at my desk for over a month with zero problems, but the sell to management failed mainly due to the cost of VCenter.
Thank you for the link to driver pack - I've been looking for something like this for a long time
VMWare could do what you want and give you multiple displays. If you do go that route make ubuntu the host and windows the guest. It works much better that way in my experience than the other way around. My guestion is are the same people using both ubuntu and Windows?
If the answer is no than virtualization is probably not what you need. Unless you are using some very very exotic hardware you should be able to put together an ubuntu image that will deploy on just about any system; with very little trouble. There are plenty of documents on doing that.
Windows Deployment Services and WAIK, will also let you build a very generic Windows Seven image; to which you can easily add the drivers for the specific platform prior to deployment. It adds it all very cleanly so you don't get the mess of registry cruft you did in the past trying to deploy a ghost image or something and fix the drivers after. Its totally different situation than it was five yeas ago if you have not kept up. We use the same core WIM image on five different model laptops for different OEMs, a couple desktops, and even a tablet with no problems.
Repeal the 17th Amendment TODAY! Also Please Read http://www.gnu.org/philosophy/right-to-read.html
Most likely I would use Citrix to deliver a desktop instead of vitualization because what you are trying to do is eliminate touching the desktop any more then you have to. VMware (or other virtual platforms) are great for running multiple "guest" on one box and can do what your looking for but Citrix has been designed ground up for delivering applications to the desktop. I would especially use Citrix if some of the machines are older and might have a hard time runnning the "host" and "guest" software. VMware (and the like) are best used for running mulitple copies of OS on one box instead of just running one standard copy. I think you would find it much easier to roll out new apps or updates to existing apps with Citrix then VMware, but in your environment it may work for you.
Why not install all the drivers for all the different machines you have into one box and then sysprep the machine. Now use clonezilla and save an image of that drive. It should now work with all the machines in the office. As far as linux on the desktop, I don't mess with that foolishness. My servers use it and it's great! My cluster uses it and it's great. My desktops ...nope.
Keep in mind that Windows 7 (and Vista) in HAL independent, and installs with Windows' new WIM imaging. This means that you can have 1 master Windows 7 image, and it will install onto ANY hardware. It's not like the "old" XP days where you needed 17 images for 17 different boxes. You can now effectively have 1 master Win7 for your entire environment, regardless of how many different kinds of hardware you have. Microsoft also has a lot of good, reasonably easy to use tools to easily maintain these .wim images. It's really easy to inject drivers into the .wim images as well, without the need for 3rd party tools (like nLite).
Down the road, as new desktops and laptops come on board, all you really have to to is update the single master image with drivers for the new box, which is usually a single download and about 5 CLI entries (on Microsoft's deployment CLI toolkit).
Don't overlook Windows 7's deployment and built-in imaging features. All of the pains of drivers and imaging from the XP days are reduced by about 98% with Windows Vista and 7.
Considering that Windows has been deployed to about a billion desktops in the world - you may want to used some MS tools to deploy (like Sysprerp) instead of CLoneZilla which isn't really intended for mixed environment deployment. Virtualization will be slower and introduce problems that will be anything but expected. KISS.....
I've run windows in KVM[0] and it works fine. I wouldn't use it for something required Direct3D or OpenGL however.
KVM runs closer to the kernel so you get better efficiency over something. If you just need windows for Office or something, you could virtualize the windows instance on the Linux box and then hook up a second hard disk and
reboot using a Linux boot CD/DVD.
Once in the Linux 'Live' boot, use dd or dd_rescue to clone the install disk to the newly attached disk. make sure
cylinders/size match on the drives. The advantage to virtualizing windows in the Linux environment, is that windows
will only see the hardware as virtulized by the host (unless you use pci passthrough). Also, using KVM, you don't
have any modules or crap (Looking at you, Vmware) when you update the kernel, or system. Also, you will have less problems
booting a cloned drive in Linux than in windows - especially if hardware is grossly different.
[0] - http://www.linux-kvm.org/page/Main_Page
boycott slashdot February 10th - 17th check out: altSlashdot.org
Citrix XenDesktop 4
http://www.citrix.com/English/ps2/products/feature.asp?contentID=1855444
Or Citrix Provisioning server
(PKA Ardent Networks)
http://www.citrix.com/English/ps2/products/product.asp?contentID=683392
Demo here http://www.youtube.com/watch?v=jAlztB6lgLU
Unattended isn't so much use if you want to put anything other than 2000/XP on your box.
Reading comprehension isn't your strong suit, is it?
Yeah that's why we are waiting for something like spice to hit mainstream:
http://www.redhat.com/virtualization/rhev/desktop/spice/
(full disclosure up front, I'm a Citrix employee)
Either Citrix XenDesktop would work for your scenario.
XenDesktop VDI edition would let you maintain a single OS image .VHD file in your datacenter. You could run it on Citrix XenServer (based on Xen.org hypervisor) for free or swap in VMware ESX/ESXi/vSphere or Microsoft Hyper-V, whatever your preference is.
Stepping up to XenDesktop Enterprise or Platinum Editions would allow you to take that single .VHD image, skip the hypervisor and stream it directly down to each different endpoint device. You would just need to prep the image to ensure it has all the neessary drivers.
XenDesktop Express is a free edition of the product that you could also use for up to 10 endpoint devices using the first delviery method. You would need to clone the .VHD image for each system as the provisioning component that allows single image management isn't available in the free edition.
http://www.citrix.com/xendesktop - Product Page
http://www.citrix.com/tryxendesktop - Free trial
Incidentally, I noticed somebody point out Citrix XenApp as a possible solution earlier. That product would be better suited to virtualize the applications not the entire desktop. It is included with both XenDesktop Enterprise and Platinum Editions as well.
It can't be done! You're crazy!
Why not just do your development on the Linux side and just test in a windows VM? You'll get the speed out of your hardware and ease of deployments.
Apart from the usual recommendations (plenty of RAM, decent graphics card, etc) run your VM from a second hard drive. I've done this successfully to run Adobe CS3 without much issue. Some people might complain that it's not as lightening fast as it is outside of a VM, but those people will always complain about any little thing they can. This is just my experience turning over an office of about 10 workstations to Ubuntu and VMware Workstation.
If you lend someone $20 and never see that person again, it was probably worth it.
Problem solved.
---- Booth was a patriot ----
Windows 7 supports booting from a VHD. You could install your PCs as Win7 booting from VHD and then back up those VHDs. If a PC failed, the VHD is portable. You also get a free XP VM via XP Mode if you are licensed for it.
This solves your problem. For every action there is an equal an opposite reaction. In this case, it is performance. Microsoft doesn't recommend doing this even though they support it.
Buy some Symantec BESR 2010 CALS or Acronis True Image & use the optional feature each has to migrate the PC to dissimilar hardware.
Image-based installation management was a flawed solution, and in heterogeneous environments, it can never be a workable long-term solution. Even virtualized workstations will drift over time, and an image-based solution won't let you fix that. To deploy a new build, you'd have to blow away everyone's customizations and make them start from scratch... that's not a good solution.
To maintain user environments while updating software loads, you still need an incremental installation setup of some kind, so why not use THAT process to build up a new workstation? Use the automated tools to script a Windows installation. Once windows is up and running, you let your software management process take over and start installing drivers and applications to the desktop. When it's all done, you have a fully-configured workstation - one that didn't rely on a pre-configured image or a virtual machine.
Yep. Before I left, we were in the talks of replacing it with WDS, since we were in the process of migrating everything to Windows 7. The general area where unattended fails is when it comes to partitioning the disk.
Someone else suggested to use OPSI; I haven't had a chance to test it out before I left, so I don't know how well it works with deploying Windows 7 nor with any sort of server OS.
"Hegelians, who love a synthesis, will probably conclude that he wears a wig." - Bertrand Russell
Simple, i am presuming that if your using clonezilla you have one or two images that your putting onto the machines, so if you only real issue with your current method is drivers for the post clone setup then all you need is to include either a large driver pack i.e driverpacks.net or create you own custom one, then get driverforge and set it to install all your hardware at first boot. Its unattended and works.
Well, I don't understand your cryptic comment, so perhaps you are right!
We build approx 300 units per year, most with the same hardware. However, Some older machines come in for repair with outdated motherboards. For this reason, when building the original Windows image, the hard drive controllers were set to generic so it will boot to any hardware. LAN drivers are already on the image, so once that's loaded, any other drivers can be installed from our PXE server (also running clonezilla). Works for us, and it's quick and efficient.
One thing I should add: it can be a little tedious to figure out what command line switches are required for whatever program you're wanting to install, especially since the whole system can install everything, including .msi's and regular .exe InstallShield packages. This tedium pays off in a big way, however: if you set it up like how the original developers set it up, where you can pick between two or three models in the bootstrap stage, then it's extremely easy for an operator to queue up 50 machines at once for installation within 10 minutes.
For example, it took me a good week or two of scripting out all of the installation stuff so that it prompts the operator for only two pieces of information: 1.) computer use type (student lab computer, laptop, etc.), and 2.) owner's username. This only takes about a minute to fill out, and streamlines the hell out of the process. Compare this to a naive Acronis image that hasn't been sysprepped, which requires about 15 minutes per computer to rename the computer, make sure that the owner has admin rights, etc. If the computer has been sysprepped, this should be reduced down to about 2 minutes, though if you're in a heterogeneous computing environment, you'll have problems down the road where you're forced to update a ton of images for however many permutations you've bothered to save to an image.
"Hegelians, who love a synthesis, will probably conclude that he wears a wig." - Bertrand Russell
In re: siblings
I suggest everyone take a break off of their official professional jobs, for about 2 weeks.
Everywhere, all at once.
Then, we can all decide whether the jobs were necessary.
Until then, the daily grind, the routine, will have clouded the judgment of the workers as to the real nature of their predicament.
That's interesting. My understanding is that applies to encrypted partitions, and only an encrypted partition not on a main hard drive.
I've found that TrueCrypt is excellent and fast at handling file-based encryption, so rather than encrypting an entire partition on a secondary hard drive, it is possible to use a file that fills that partition.
I understand better now.
I think this is an important point: You can make a TrueCrypt-encrypted file that is almost as large as the USB drive capacity. There seems to be no performance loss when doing that.
Then you can put a copy of TrueCrypt, unencrypted on the USB drive, also, and any other files you may not want to encrypt.
I use a 4 GB TrueCrypt-encrypted file that is a little smaller than the space on a blank DVD. That facilitates easy backups. The encrypted file on the DVD can be mounted and read also; that works very well.
In xp, basically it seems you have to get the HAL model adjusted for each machine, then the individual drivers. I haven't found my solution yet, but I did find these. -- UIU - Universal Imaging Utility - www.uiu4you.com -- acronis universal restore -- sysprep - driverpacks.net/
Build your own energy sources from scratch. http://otherpower.com/
Virtual Bridges / Win4lin seem to have some sort of windows plugin for linux, but I can't quite remember how it works. IBM seems to have liked it and partnered with them. Anyway, that would be a nice product, some sort super-stripped down windows-compatibility-library, that runs windows legacy apps, and end of story, no more giving windows any partitions or hardware control.
Build your own energy sources from scratch. http://otherpower.com/
A company I worked for got around this by ordering all their hardware from dell and always specifying the same hardware so the drivers would be the same. The configured the first system to work the way they wanted with their MS server and then cloned the working drive for all the other workstations. All data was stored on the server. suspect an infected machine, reclone the drive.
We use a product by Binary Research called Universal Imaging Utility (http://www.binaryresearch.net/products/the-universal-imaging-utility) for Windows machines. It basically bundles a ton of drivers into the sysprep process and lets you use a single image for disparate hardware. It costs about $20 per machine.