The case for Thin Clients

The case against thin clients is that fat clients are so inexpensive that for little or no extra money, why not get a fat client instead of a thin one? Nowadays, it costs more to deploy thin clients and a fat server than it does to put fat clients on every desktop and a central file server (Network Attached Storage device).

The prices of some manufacturers' thin clients are higher than many of their fat clients. For example, the Dell OptiPlex FX170 thin client device with Windows Embedded Standard 2009 sells for $550 (Intel Atom N270, 1 GB RAM, 1 GB Flash storage, no hard disk, no monitor). Dell sells a Vostro 15" laptop running Windows 7 Pro for $100 less money (Intel i3, 2GB RAM, 250GB HDD)!?

It is hard to argue that there is greater value in a thin client than a fat client, so most of our clients continue to replace their old PCs with new machines running Microsoft Windows and Office. But the case is not always black and white.

Service calls on broken PCs can cost hundreds of dollars.  Properly factored into the decision, this can tip the value proposition in favor of thin clients.  One of the virtues of thin clients is that they are rock solid. No moving parts. No viruses. Very little that users can do to break them (particularly if the thin clients are not running embedded Windows).

Some may argue that there are things one can do to reduce the maintenance and repair costs on fat clients, but these things are not "user friendly."  

• Restrict what users can do that might cause problems, such as (sharply) limiting users' rights or forcing them to use "mandatory profiles".  Setting up user restrictions can be quite time consuming and costly.  
• When problems occur, restore or replace the PC with a standard, working image or PC.  In other words, do little or no troubleshooting and problem-solving.  This is a policy that is hard to hew to in the real world.

The case for thin clients hinges on the following:

• Using low-cost thin clients, and avoiding embedded MS Windows.  We are deploying $200 HP t5335z devices at one client location.
• Utilizing the total cost of ownership over the life cycle of the competing devices to decide which to deploy.
• Ignoring the preferences of users if they would rather have fat clients (most would).

Clone a failing Windows hard disk with ddrescue on Ubuntu Rescue Remix

Scenario: The hard disk is starting to throw "bad block" errors. Windows still works/boots, but who knows for how much longer. And you don't want to have to go through the time and expense of rescuing the data from the failing disk, installing Windows and all your applications on a new hard disk, and re-loading the data on the new hard disk.

Solution: You want to use a free Linux tool called Gnu ddrescue to clone the failing drive to the replacement drive.  This is one of the tools available on the Ubuntu Rescue Remix.  For more info on this tool, start here: http://www.gnu.org/s/ddrescue/ddrescue.html

If you Google ddrescue, you'll find lots of info on how to use ddrescue to recover data.  However, there's not a lot on how to move a Windows load from a failing drive to a new drive.  For example, one mistake that I've made is imaging the failing drive instead of cloning it.  The image file is not easily restored and bootable on the new drive. 

Here's what you need to do.  Idiots beware, this is not an idiot-proof guide.  Important but obvious steps are omitted.  Feel free to embellish or enumerate such steps in your comments.

1. Download Ubuntu Rescue Remix here: http://ubuntu-rescue-remix.org/Download and burn it to a CD.
2. Get a hard disk to replace the failing drive that is the same size or larger.
3. Connect the new disk to the machine with the failing disk.  A fast connection (eSATA) is better than a slow connection (USB).
4. Boot the machine to the Ubuntu Rescue Remix CD.  This will get you to a command-line prompt; there is no GUI.
5. At the command prompt, key in the command:
   sudo lshw -C Disk -short .  This command will identify the disks as /dev/sda and /dev/sdb (or perhaps /dev/hda and /dev/hdb, or perhaps you have more than 2 drives connected, ...).  This command will list the size and make of each drive, which should allow you to determine which drive designation is the source/failing drive and which is the target/replacement drive. You need to know this without ambiguity; it is easier if the target is larger than the source.  
6. For the sake of example, let's assume that the source drive designation is determined to be /dev/sda and the target drive designation is /dev/sdb.  Your situation may be different.  To clone the source drive, at the command prompt, key in the command:
   sudo ddrescue -v -r 3 /dev/sda /dev/sdb logfile
   That's it.  Depending on the size of the source drive and the number of bad blocks, it will take time to clone the drive.
7. When ddrescue is done, swap out the target disk for the source disk in the PC, and fire it up.  It should boot and work as well as the source disk, but without the looming prospect of imminent disk failure.
8. If you have used a larger target disk to clone the source, you may want to expand the partition to use the whole, larger disk.  GParted will do the trick.  Unfortuantely, this tool is not on the Ubuntu Rescue Remix CD.  Parted is, but it will not resize NTFS partitions.  You can find GParted on the Ubuntu live CD.  You can download it here:  http://www.ubuntu.com/download/ubuntu/download.  Instructions for using GParted to resize a disk are here: http://gparted.sourceforge.net/larry/resize/resizing.htm.