NAS Devices Come To The Fore.

CD/DVD update

Optical storage has seen some radical changes over the years that have made it a more commercially viable storage option. In general I will focus on the CD and DVD media when referring to optical media. Like any technology there continues to be advances in the industry that change the way it is used and what options the new technology provides. Often the jumps in technology occur and then people invent new and expanded uses for the technology that were not previously considered.

CD media has been the standard software distribution method for a number of years now. The high capacity and low cost of the media made it the logical replacement to the floppy. The rapid growth in program size led to this development. As programs began to take up 30 and 40 diskettes it became economically feasible to switch to CD as the medium for software distribution. This led to a rapid increase in the needed units of both read only devices and recording devices. The increased unit sales spurred lower unit cost, which increased the demand until it eventually became inexpensive enough that they were a viable product for the consumer.

CD media became commercially used because it was inexpensive enough for content providers to use and therefore there was a rapid expansion in the amount of content available. As is often found in other storage media, increased capacity leads to increased usage. Companies quickly found that they were often getting data on multiple disks, with some data intensive applications coming with as many as 20 disks. At this point single disk capacity drives became a bottleneck, which led to the development of disk servers.

There are three basic types of multi-disk servers: direct attached towers, jukeboxes, and Network Attached Storage (NAS) devices. The direct attached servers are the brute force method of solving the solution, simply put they are large towers with banks of drives attached to a server; by putting many drives in the system you could have access to many CDs at once. There are some drawbacks to this system. You have a highly inefficient use of drives, which mostly sit unused while their disks are not being accessed. This also required management software to allow all the disks to be managed, and also put additional burden on the server. Finally these systems are slower than hard drive based units since reading from a CD is slower than data access from a hard drive, and users noticed this slower performance.

The logical extension of the brute force method was the jukebox. This made more efficient use of the read drives since only the data being accessed is placed in a read drive; the other disks are stored for later retrieval. The drawback here was that, just like the jukeboxes so popular in restaurants and bars, you have to wait while the machine finds, gets and places the disk in the drive for reading. These systems gained in efficiency but the wait time for data access increased considerably as you now had robotic wait time added to the relatively slow access time. The jukeboxes did however allow huge disk capacity: while they are relatively expensive it was more efficient and economical than the equivalent drive bank method to reach the same number of disks accessible.

The next class of products to enter the market were the NAS type devices. These come in two varieties, caching and non-caching. Caching is the process of placing the contents of the CD onto a hard drive to allow access to the data at hard drive speeds as opposed to CD-ROM speeds. The non-caching devices are generally small and do not provide large numbers of online disk capability. The caching devices are where technology advances have made a real difference. Rapidly increasing hard drive capacities along with the decreasing cost per megabyte have made these devices very attractive. For example, Procom's mid-range Dataforce product line offers 255 disk capacity in a desktop form factor under $10,000 dollars, while jukeboxes in this capacity range cost on average between $14,000 and $18,000 dollars. Disk capacities are only expected to grow and make these mid and low range products more competitive with the jukebox products, while avoiding all of the performance issues associated with the optical media read r ate and the robotics.

This presents a less than optimistic outlook for the future of jukebox sales. Industry data has shown that the only real growth sector of the jukebox market has been in the over-200 slot units, while sales in other sectors have been flat or declining. With performance and price pressure from the NAS devices it is likely that the high-end jukeboxes will start to follow the same pattern as the lower end devices.

Jukeboxes face a number of problems, related to disk capacity, which potentially will hurt their future sales. In order to increase disk capacity in a jukebox, the physical size of the unit must be increased, which increases the cost of the unit. Jukebox manufacturers have to be very careful to maintain a constant cost per disk stored ratio, as the increased size means not only a larger enclosure but also a large travel distance for the robotics involved. A direct result of the larger capacity and larger travel is that the peripheral disks are farther from the read drives which could increase travel time and thus lower overall unit performance. This can obviously be overcome with more advanced, faster robotics but these also increase the cost of the unit and require additional R&D expenses that the company must make and recover by passing the cost along to its customers. With larger numbers of disks the likelihood that the read drive capacity will become a bottleneck increases. Overcoming this issue requires the addition of more read devices and also increasing the physical dimensions of the unit, both of which add cost as mentioned previously. Finally, all of these problems are overshadowed by the fact that most companies are space conscious and are trying to reduce the overall space used up in their data centers, or they are trying to reserve as much space as possible for their growing need for hard disk based storage.

The caching systems have a distinct advantage in the disk capacity game since the growth is contained within the hard drive. Additionally, the cost of the R&D is borne by the drive manufacturer, who passes the cost along in the normal course of business, but over a much larger unit sales base. This means that the caching systems are more likely to continue to be able to provide a decreasing cost per disk stored, while also not paying a penalty in performance for the increased storage capacity. The one issue to be wary of is that, in caching systems, if a hard drive fails, the disks must be re-cached, which can be time consuming.

The next major change affecting this market will be the adoption of DVD as the primary source for data distribution and storage. Up until now the cost of purchasing DVD writing units has been very prohibitive for most companies and the capacity has not been needed. Manufacturers of the read devices have continued to push forward and have now developed units that are cost effective enough to become the new standard in the PC market, both mobile and desktop. With an increasing potential user base it is becoming feasible for companies to start providing content on DVD. The entertainment industry has been the pioneer of the DVD content market since it lent itself well to providing movies to the general public. While this has provided a strong consumer market for the reader manufacturers, there has been little progress in terms of business content growth, Companies have, however, been able to build DVD read capacity with no additional incurred costs and are now poised for the introduction of DVD content.

Microsoft's planned release of future software on DVD media is likely to be the first major business oriented content provided. Additionally, a number of manufacturers have indicated that mid 2001 we could expect to see lower cost DVD writing units available. This, coupled with the falling cost of DVD-R material, may entice many content providers to switch to DVD and take advantage of the increased storage capacity available.

The interesting dynamic will be the race between jukeboxes' ability to provide large DVD access at more attractive pricing during the time frame it takes caching devices to increase drive capacity, to accommodate the larger amount of data on DVDs.

We have seen many changes in the optical storage arena and are poised to see the conversion to DVD storage in the next year or so.

Kevin Judd is the workgroup NAS product manager at Procom Technology (Irvine, CA).