Build a Virtual CD-ROM Jukebox

This article describes how to set up a virtual CD-ROM jukebox (VCDJ) using Samba and Linux. A VCDJ is a network server that provides access to the contents of a large number of CD-ROM disks, without the need for more than one CD-ROM drive. In addition, it simultaneously provides access to the ISO 9660 CD-ROM images in a format suitable for burning copies of the CD-ROMs using a CD-RW drive.

A CD-ROM jukebox is usually a file server (or file server appliance) connected to a CD-ROM drive tower. It is able to share (often via SMB/Windows Networking) the contents of a number of CD-ROMs to clients on the network. It's valuable because users of the network do not have to locate a particular CD-ROM physically when they wish to install software or access data.

However, this approach suffers from some drawbacks. The number of CD-ROMs it can serve is limited to the number of CD-ROM drives in its tower(s). To add more CD-ROMs, more drives must be obtained and installed. The CD-ROMs must be in the drives at all times, making them unavailable for other purposes. Also, there is no easy way to make copies of the CD-ROMs (especially bootable copies) without removing them from the server, which makes them unavailable for network users.

A VCDJ surmounts all of these limitations. It is different from a regular file server because, while a regular file server might contain the contents of a number of CD-ROMs, the VCDJ contains an ISO 9660 image of the CD-ROMs. When we're done, it will serve both the images and the contents of the images efficiently (on a file-by-file basis) at the same time. Additionally, the original CD-ROM disks can be stored away where they won't get lost.

Whereas traditional CD-ROM servers are limited by the amount of CD-ROM drives they contain, the VCDJ is limited by the amount of disk space it contains. Hard drives are an order of magnitude cheaper than CD-ROM towers, and they scale better. One 40GB hard drive occupies one spot on an IDE or SCSI controller. It can contain the equivalent of 57 full-sized CD-ROMs (at 700MB each). We would need 57 CD-ROM drives attached to the server to get the equivalent functionality, which is a practical impossibility.

At my place of work, we have found the VCDJ invaluable for publishing the contents of regularly updated software subscriptions. We used to lose track of CD-ROM disks as we loaned them out to others; now we just give them access via their Windows Domain credentials. And, we easily can burn new copies of any bootable CD-ROMs we need. The original disks remain locked away.

To create our VCDJ, we'll need the following pieces:

The first task is to obtain the ISO 9660 images of the CD-ROM disks. Any tool you can use to, say, make duplicates of CD-ROMs can generate proper images. You also can download ISO images of your favorite Linux distribution.

On Linux, the simplest way to make an image is with cat. Put the desired CD-ROM disk into the CD-ROM drive. Make sure the directory /mnt/images/ exists. If your CD-ROM disk block device is hdc, the image is created like this:

cat /dev/hdc > /mnt/images/image1.iso

You'll want to give the image file a more descriptive name. Reading the image may take awhile. Repeat this process for each CD-ROM disk of which you want an image.

Now that we have the CD-ROM images, we'd like to access the contents of the images. The normal method for accessing the contents of the image is to use a loopback device, like this:

mount -t iso9660 -o loop,ro /mnt/images/image1.iso
        /mnt/isosrv/image1/

This mount command says that we are going to mount some data that uses the ISO 9660 filesystem format. It also says to use the loopback device. The loopback device is a nifty kernel feature that allows you to designate a file, in this case /mnt/images/image1.iso, to be used as if it were a character device, like a hard disk or CD-ROM drive. This command mounts the image file in a read-only format. The contents of the CD-ROM image can be seen in /mnt/isosrv/image1/.

How Do We Know That the Image Is Correct?