Jukeboxes make it easy to store a lot of data. They're cheaper than RAID systems. Jukeboxes can have hundreds of discs and dozens of drives. Their robotics swap discs quickly and easily. But you need software to drive the jukeboxes and to manage the discs and files stored on them.
Good jukebox management software (jukeware) keeps your jukeboxes humming along. It catalogs files. It operates the jukebox. It identifies and tracks the media in the jukebox. Sometimes it tracks your offline media. It caches files saved to a jukebox. It mounts and unmounts discs. It logs in new discs and knows when disks are removed. In its spare time, it optimizes itself and the jukebox for maximum speed.
Since some jukeboxes hold more than a terabyte (a thousand gigabytes!), keeping track of files is a massive job. Jukeware creates a database linking the file information needed by the operating system to the location and ID of the disc where the file is physically stored. When you request a file, the jukebox management software checks its database for the file's location. The jukebox is then commanded to move the disc containing the file to a disc drive.
This lets the jukebox connect to the computer as if it were a single physical volume. The single physical volume can be partitioned into two or more logical volumes. A logical volume can be a single disc or can span several discs. This allows great flexibility in controlling how a jukebox is accessed and used while maintaining transparent operations for end users.
Some document management applications keep track of documents with their own databases. When you use a document manager with typical jukebox management software you end up with two databases, one for the document manager and another for the jukebox manager.
Kofax (Irvine, CA 714-727-1733) uses ActiveX controls to add jukebox management directly to an imaging application. This lets developers use the database built into imaging and document management applications to keep track of documents on a jukebox. Storage Controls ($1,000) is an ActiveX toolkit that lets VARs add jukebox management to any document management system.
This approach improves performance in a single document management system. The advantage is especially marked on large jukeboxes with huge numbers of files. Storage Controls integrate the jukebox's operations tightly with your imaging application. The downside is a loss of flexibility. You have to integrate new software with the imaging application to use the jukebox.
While jukeboxes work with standard media, there's nothing standard about jukeboxes. There are dozens of different jukeboxes from as many different manufacturers. If you've ever been to an AIIM show, you know how many ways there are to feed a CD into a drive.
CD-ROM jukeboxes are simple. CDs are single-sided. To operate a CD-ROM jukebox, the jukeware only needs to keep track of discs, drives and their locations. Some jukeboxes have a single drive. These are the simplest. If the CD jukebox has a CD-R drive in it, the software's job becomes more complex. It must make sure that the CD-R drive is reserved for writing data.
Magneto-optical jukeboxes don't need to keep track of which drives can and can't record. Magneto-optical drives do both. Even though magneto-optical discs are double-sided, the drives are single-sided. This means the jukeware has to keep track of both sides of each disc. When a storage technology is new, there's usually a single media type. For a long time, CD-jukeware had only readable media and a single data density to worry about. Even when CD-R came out, the new media was write once. Once a CD-R was burned, it could be treated like any other CD-ROM.
New CD technology will complicate matters. Rewritable CDs aren't compatible with most older CD-ROM drives. DVDs, because their form factor is similar to CDs, will have to coexist with CD technology. DVDs come in several different formats. The discs are double-sided. They can be single- or double-layered. The DVD industry is working on recordable DVD and rewritable DVD. Jukeware vendors are working hard to keep their products up-to-date. They realize the need to keep them compatible with new technologies as they become available.
Magneto-optical discs have a number of different data densities. These range from 650 megabytes to more than 4 1/2 gigabytes. Magneto-optical drives are downward compatible. Older 650 megabyte discs work in the new high-density drives. Jukeware has to keep track of how much free space is available on each disc. If a user saves a file, particularly a large file, the jukeware locates a disc with enough free space to store the file. Mistakes cost time and flexibility.
The Universal Disc Format (UDF) may simplify media handling in the near future. UDF is a file format originally devised for DVD. It's now being adapted to other optical media. If UDF catches on, it could be the one file format all software and hardware manufacturers agree on for all disc types. UDF is slowly being deployed on CD-R and CD-RW. When DVD is in full force, UDF will grow.
The plethora of CD formats is handled by Smart Storage's (Andover, MA 508-623-3300) Smart CD ($500-$10,000). Smart CD manages jukebox reading and writing. With Smart CD you can record to CD-R across a network. It has intelligent caching on both the read and write side. Information about the CDs within a jukebox is stored in a proprietary-optimized database. Smart CD gives single drive letter access to a jukebox. Ease of use was a primary consideration in Smart CD's design.
Smart Storage is building packet writing technology and UDF into their products. UDF support can be used on magneto-optical and WORM media.
Jukeware mounts and unmounts discs. When you first start up a jukebox, the software polls all the discs in the jukebox and builds a database of all files. This takes from 30 minutes to several hours depending on the software and the size of the jukebox.
Although setup is only done once, the time it takes for a jukeware package to set up its database is a good indication of how it will perform in day-to-day operations. Jukeware that takes a long time to setup may have a slow database engine. It may also have problems scaling to larger jukeboxes or to large numbers of users. Once the jukeware is set up, it needs to add and remove discs. Good jukeware lets you add and remove discs without taking the jukebox or the software offline.
Micro Design International (Winter Park, FL 407-677-8333) sells a number of jukeboxes for CD-ROM, magneto-optical and Write Once Read Many (WORM) optical drives. To support their jukeboxes they make SCSI Express jukeware. SCSI Express is also available as a shrink-wrapped jukebox management solution ($4,000-$9,000).
SCSI Express' designers make a point of making the jukeware easy to manage. Media detection is automatic. Discs on the jukeboxes can be grouped into logical volumes. You can even save files across several discs. This is useful for very large files. SCSI Express avoids the pitfalls of striping files across several discs. It requires all the discs to be mounted in a jukebox before a volume is accessible.
SCSI Express maintains performance with multi-threading and by maintaining read ahead and write back caches. Micro Design maintains compatibility with older and smaller systems with a Netware version of their product. You can use SCSI Express with any jukebox, but it gives enhanced performance with Micro Design jukeboxes.
Optimize for Performance
Every jukebox manufacturer brags about the speed of their jukeboxes. When you buy a jukebox, look for minimum swap time and drive-to-disc ratio. Jukeware has to support the performance features built into jukeboxes.
Multi-threading is when jukeware simultaneously processes multiple requests. The advantages are most noticeable when you conduct jukebox housekeeping. A multi-threaded jukebox management system keeps the jukebox running and serving users while you add and remove discs. The jukeware serves file requests while it updates its database. Multi-threading is important when you run multiple jukeboxes off the same server.
Intelligent disc queueing prevents disc thrashing. Thrashing occurs when a jukebox receives multiple simultaneous file requests. If a jukebox serves files in the order the requests are received it can end up spending most of its time swapping discs. This is especially bad when the requested files are small. Drives can spend a lot of time sitting idle while the jukebox robotics swap discs.
An intelligent disc queuing mechanism looks ahead in the file request queue and groups the requests made to each disc together. If a large number of requests are made to a single disc, intelligent jukeware makes sure one disc serves all the requests at one time. Disc swapping is minimized. The jukebox is used at its maximum efficiency.
Jukeware makes the jukebox work as efficiently as possible with file caching. When a file is requested, the next file will probably also be requested. Jukeware can "guess" which file will be requested next and store it in memory. The drive can then be freed for the next request. If the jukeware guesses right, it saves a disc swap.
Caching is the most important feature in Disc Extender ($400+) from OTG Software (Bethesda, MD 301-897-1400). Disc Extender has both read and write caches. Retrieved files are automatically stored in a read cache in case the file is requested more than once or by more than one user. A write cache is kept for all saved files. The files are saved to disc when user activity slows.
Disc Extender gives drive letter access to any media on any number of jukeboxes. Discs can be grouped within a jukebox or across more than one jukebox. Just about all media is supported (WORM, CD-R, magneto-optical and tape). Disc Extender has a full set of tools for administration. It tracks files on a jukebox using standard database software. Either Sequel or Oracle Personal Edition database software can be used. Standard database software can easily be incorporated into a document imaging system.
Optimization features are user-adjustable. To take best advantage of them, jukeware should keep a log of file and disc requests. If the jukeware doesn't automatically optimize itself, use the log information to optimize it manually. Even jukeware packages with automatic optimization run into situations where they work less efficiently. User adjustable optimization lets you tailor your jukebox management to your application. The jukeware application should provide the tools needed to optimize it properly.
There are many jukeware packages for Unix. EMASS' (Garland, TX 972-205-5665) AMASS ($4,000-$70,000) supports a wide variety of jukeboxes under Unix. AMASS also manages tape libraries. AMASS could be the software package to use if you need one application that manages all your jukeboxes and libraries.
Computer Associates (Islandia, NY 516-342-5224) makes three jukebox management packages: CD Server ($1,500) for CD-ROM jukeboxes, and Optical Server ($4,500) and File Jockey ($4,000) for magneto-optical and WORM jukeboxes. CA stresses ease of use and installation. Both File Jockey and Optical Server offer drive letter access to the jukebox. Optical Server lets discs be grouped together. File Jockey only makes each disc available as a separate directory. CA's jukeware keeps track of discs outside of the jukebox.
Luminex's (Riverside, CA 909-781-4100) Fire Series jukebox management software runs on all the popular Unix platforms. It reads and writes to any CD format. This is a good choice for multi-platform environments. Every computer on the network can read any CD in the jukebox -- even if the CD is written in a foreign format.
Fire Series for Unix handles the compatibility issues. The Fire Series software comes bundled with Luminex's jukeboxes with prices starting at $13,000. Fire Series has a full set of features for performance enhancement. It includes permanent and temporary caching and it is multi-threaded. If Unix is not your thing, Luminex has a version of Fire Series for Novell.
KOM (Ottawa, Canada 613-599-7205) uses the native file system built into your OS to track files within a jukebox. This keeps the jukeware non-proprietary and compatible with multiple operating systems. KOM's jukeware is compatible with all server operating systems: OptiStorm on the Windows NT side and OptiServer under Unix ($750-$75,000). Both software packages are available in a variety of configurations for use with any jukebox. Or use one of KOM's jukeboxes.
For their own and other manufacturers' jukeboxes, Mitsubishi Chemical America (Sunnyvale, CA 408-773-2836) has JukeAid ($4,000-$7,000). JukeAid is Novell certified magneto-optical jukebox access. It turns an MO jukebox into a Netware volume. A Windows NT version of JukeAid is forthcoming. JukeAid is multi-threaded. It has patented write cache technology that lets files be saved across multiple discs.
Jukebox optimization takes many forms. The most common is caching. The second most common is to physically locate frequently used discs as close to the drives within a jukebox. Tower Technology (Boston, MA 617-236-5500) goes one step further with the management of WORM drives.
Their clustering technique writes to discs evenly throughout the jukebox instead of filling up one disc and moving to the next. Related documents are kept on a single disc whenever possible. This saves the jukebox disc swaps.
This optimization is only possible when the document management system directly controls the jukebox. Tower's jukebox management software is part of the Tower Imaging System ($100,000).
Plasmon's (Minneapolis, MN 612-946-4100) jukebox management software is compatible with hardware from most vendors as well as their own jukeboxes. Multitasking Manager ($750-$20,600) supports magneto-optical and WORM jukeboxes on Unix, Windows NT and OS/2 platforms. It uses multi-threading and an optimized proprietary database to ensure fast performance.
Tracer Technologies (Gaithersburg, MD 301-721-2800) makes two jukebox management products. MagnaVault ($3,500-$30,000) is designed for magneto-optical and WORM media. HyperROM ($2,400-$9,000) is for CD-ROM. Both products work under Unix.
They have built-in databases and prefetching read-ahead cache. When you request one file with Tracer jukeware, the software reads ahead to the next file. Prefetching speeds jukebox performance and reduces time-consuming robot movements.
Ornetix (San Jose, CA 408-383-7050) wants your jukebox on a network. Their multitasking and multithreaded CD-Vision jukebox software ($1,650-$7,000) makes sure this happens. CD-Commander, a separate application included with CD-Vision, administers and manages jukeboxes. CD-Vision lets you use a 386 PC as a jukebox server on Novel or Windows NT networks.
Some of the heaviest users of CD-ROM resources are publishers and output bureaus. Many use Macintoshes. It's hard to find jukeware for operating systems other than Unix and Windows NT. One of the exceptions is Miles Apart ($3.600-$4,300) from Astarte (Karlsruhe, Germany 011-49-721-985-540). Miles Apart provides all the usual jukeware features, multithreading, multitasking and caching for Macintosh only environments like publishing and multimedia shops.
