Enterprise Storage Solutions For Data & Document Management Enterprise storage management: The past and the present By Michael Johnson, Business Development Manager Tracer Technologies Inc. During the past 20 years, the migration from mainframe computing to highly distributed client-server models has created challenging data storage management issues. Much attention has been focused on open systems as interoperability between decentralized storage servers has become a paramount goal in the management of an enterprise's most important asset-the data. As distributed client-server networks proliferate and their role in handling mission-critical applications and data grows, so does the need for fault tolerance and redundancy to minimize equipment failure and reduce data loss. Equally vital is the need for increased data and file accessibility, security, version control, long-term archival storage and disaster recovery. According to several industry analysts, the average Fortune 500 company manages over a terabyte of data on a daily basis. By the new millenium, company data storage numbers are expected to grow to over a petabyte, or 1 billion megabytes, averaging a 57% growth rate annually. Data from several research firms reveals that the average Unix site will add an additional 40GB of disk storage annually, while the average PC site will add 90GB per site. The cost of storing and administering such large data storage environments, averaging between $300,000 to $700,000 annually per network, has driven the demand for increased densities of storage peripherals equally. Additional productivity and time-to-market demands have also increased the need to analyze old data to help organizations perform cost analyses based on past successes and failures quickly and efficiently. Along with decentralization comes fragmentation of data and databases across multiple platforms in the corporate enterprise. Corporate data storage server configurations are quickly maximized, increasing the need and demand for high-speed data access, security and bandwidth over the network. Attaching the storage server directly to the network has become the latest trend in handling increased LAN performance and fault tolerance demands as users share files across different computing platforms. This new concept of network-ready storage (NRS) provides both off-loading of CPU resources and storage requirements from a dedicated server while offering increased data accessibility to a heterogeneous user environment. Data warehousing and data mining from these SQL-accessible decentralized storage servers bring us full-circle back to the need for a centralized RDBMS storage pool that encompasses all storage technologies and strategies available in the storage hierarchy of both today and tomorrow. Coupled with a powerful HTML-based management front end, NRS-centric solutions will be in great demand into the new millenium. With the reduction in cost of ownership for automated on-line and near-line (jukeboxes) storage peripherals, most companies will opt to keep their data on-line. One of the shortcomings of an off-line storage strategy utilizing traditional backup/restore technology is that although it is good for an incremental save/restore, full restoration from a disaster recovery perspective for terabytes of data becomes extremely lengthy, cumbersome, and in most cases, an impossible ordeal. Although tape has been generally accepted as the backup storage medium of choice from a cost perspective (averaging $.02 per megabyte), severe limitations are inherent in accessing data from a non-random access (sequential) storage medium. Proprietary data formats have also caused data interchange problems for tape users between the disparate distributed network storage clients. The System Independent Data Format (SIDF) standard for tape data interchange is one proposed standard that resolves both the data format and ability to mount the saveset issues, but still has major supported platform limitations. New product offerings in 1997 will bring on-line mountable tape file systems that will provide quick access to previously backed-up data and file systems while offering a variety of standard and unique file indexing formats. Although magnetic disk provides random access on average for $.20 per megabyte, it too quickly becomes cost-prohibitive when factoring 100% data redundancy and both removable and archival storage requirements. Whether utilizing storage array or RAID technologies, data integrity, format and overall file structures are the biggest concerns for the enterprise storage manager, along with the inability to perform complete daily backups to such large multigigabyte storage subsystems. Increased storage capacities and access times, multifunction capabilities and a plethora of hardware and software interchange standards have propelled optical storage technologies to the forefront for both on-line and archival data storage requirements. CD-ROM continues to lead with ISO-9660 compatibility on virtually any host platform and operating system environment. With the recent introduction of CD-ROM jukebox products, CD-ROM has quickly become the most cost-effective on-line archival retrieval subsystem available. With the new high-density digital video disc (DVD) and MCD products coming to market in early 1997, CD-ROM will remain an integral component in the storage hierarchy into the next century. The availability of standard formats coupled with a price of approximately $.07 per megabyte is causing CD-Recordable (CD-R) to gain marketshare quickly for both archival data storage and backup/restore save sets. Although CD-R provides the appearance of random access to data once it has been recorded, the painful staging and sequential recording process limits CD-R's practical on-line application for large archival storage and backup requirements in most traditional enterprise client-server environments. However, tightly integrated into an automated library storage environment and coupled with the upcoming Universal Disc Format (UDF) data format, CD-Recordable technology offers promise for the future as the on-line archival storage and backup/restore medium of choice. Both 3.5- and 5.25-in. optical storage technologies are ideal for intermediate and long-term data storage, be it on-line or archival. Offering both removability and random access capabilities, these form factors provide many international standards for hardware, media and software interface capabilities. While 12-in. technology may be superior in large terabyte imaging, for video and archival storage environments that require many gigabytes of data underhead it offers no international standards, which limits its mass appeal. However, the 12-in. form factor is still a viable solution for several large niche markets that require storage of single large files, that is, less than 2GB, that would require caching, spanning and continuous swapping in a 5.25-in. form factor storage subsystem. However, with the advent of 5.25-in. storage capacities of 4.6GB, 5.2GB and higher, the non-standard and limited sourcing 12-in. form factor becomes a less-attractive solution in a long-term archival storage strategy. An approach for the future Interfacing these varied optical storage technologies into a wide variety of host platforms and operating systems has been a niche for many companies over the past 10 years. Some optical storage management vendors offer only one of several approaches, including a file migration scheme, a hard disk emulation, or a magnetic disk-based database implementation. Each of these approaches has limitations and none of them differentiates between storage methods for write-once and rewritable media. In a file migration scheme where the optical is treated as the backup storage for archiving, many of these systems bog down during data swapping, when the statistical assumptions that govern their design do not work for imaging and large-scale scientific applications. Treating the optical media as a collection of magnetic hard disks works well on a single platter but causes massive platter-swapping and thrashing in a jukebox environment because the rest of the system and application is not "jukebox-aware." Systems that store data only on the optical platter and keep directory and index attributes in a magnetic disk-based database become unmanageable when it is necessary to use multiple servers on a network, transport optical media to another server, and when off-site backups are required. Treating the optical media as the primary storage device offers several advantages. These include instantaneous, on-line access to files and data without waiting for file migration back to a client or server; self-describing media providing data security and interchangeability; and unique rewritable and write-once file systems appearing as standard magnetic disk file systems to your applications and users while taking full advantage of the unique properties of each technology. Rewritable media are managed in a way that is similar to hard disk management, except that allocation, storage and retrieval methods are "jukebox-aware." One example is that space is automatically recovered when a file is deleted, just as it is on a hard disk. Data management techniques also place data in such a way to avoid thrashing while providing maximum optical drive performance and throughput. No "dead space" or "file system grooming" utilities are required to remove previously mapped files because a write-once file system was used to manage rewritable media. Tracer's write-once file system offers an audit trail to file changes, even though the write-once file system provides the illusion of a fully rewritable file system, while still providing the ability to retrieve historical copies of deleted or changed files. Tracer includes an HTML-based on-line, interactive documentation feature that provides both intuitive, interactive on-line help and optical subsystem management from a Web browser. Also included is a migration tool kit for automatic migration of magnetic file systems to optical file systems, Migration Toolkit offers on-line access to mounted optical file systems without having to migrate back to a magnetic disk environment for client access. Future products will include a direct network-attach storage server product that includes both an administrative and user HTML interface, server-side migration, spannable rewritable and write-once file systems and CD-Recording capabilities coupled with NFS, LanManager, SPX/IPX and AppleTalk network protocol stacks. Tracer's product offerings support the DEC OSF/1, DEC Unix, HP-UX, IBM AIX, SGI IRIX, SunOS, Sun Solaris 1, Solaris 2 and Solaris x86, and now Windows NT host platforms. Michael Johnson is the business development manager at Tracer Technologies Inc. (Gaithersburg, MD), 800-319-7348, fax 909-780-5708, E-mail Mpj1@aol.com.
	TOP OF PAGE		HOME ++ SEARCH IW ++ DAILY NEWSWIRE ++ CURRENT PUBLICATIONS ++ ABOUT IW BUYER'S GUIDE ++ DIIME ++ SUBSCRIBE ++ COMMENTS
	© 1995, 1996, 1997 Cardinal Business Media, Inc. All Rights Reserved. The names, logos and icons identifying CBMÆs products and services are proprietary marks of Cardinal Business Media, Inc. CBM has no liability for content or goods on the Internet except as set forth in the Terms and Conditions of Service.