About EMC



Advanced Search

Company Information

Management Team

Vision

History

 


A Brief History of EMC through 1998

From "EMC Corporation: From Inception through 1998 - The Rise to Market Leadership," Babson College, Case 399-161-1.

Memory Products

EMC Corporation was founded by Richard J. Egan and Roger Marino in Newton, Massachusetts in August 1979. Egan, a former Intel General Manager, was attracted by the opportunity to sell add-on memory to the rapidly growing minicomputer marketplace. EMC’s first product was a 64 kilobyte chip memory board developed for Prime Computer and introduced in 1981. Sales passed the $3 million mark in 1982 and reached $18.8 million two years later. In the mid-1980’s EMC demonstrated its commitment to technical innovation with a series of memory and storage products that improved performance and capacity for minicomputers made by IBM, Hewlett-Packard, Wang, and Digital Equipment. Offerings included expansion cards and enhancement boards for main memory, storage controller units, communications controllers, and multi-function units.

The company went public in April 1986, a year in which sales hit $66.6 million and net income $18.6 million. In the late 1980’s product offerings extended from single components to complete sub-systems, such as conversion kits that enabled end users to replace entire main memory sub-systems - boards, controllers, backplane, power supply and interface modules - to enhance capacity and performance. EMC also expanded strongly into the auxiliary storage arena, where it remarketed other suppliers’ magnetic disk drive storage subsystems, often coupled with its own controller units, and also ventured into optical laser disk storage. In 1987 the company introduced solid state disk (SSD) storage systems for the mini-computer market. Consisting of controllers, which emulated storage directors, and memory cards, which functioned like rotational disk drives, solid state devices offered 20 times the performance of magnetic disk drives. Overly aggressive product line expansion, defective components, industry-wide memory price declines and end of product life-cycles all contributed to decreased revenue and a net loss of $7.5 million for 1988. At the annual shareholders meeting in May 1989, founders Egan and Marino pledged to forgo pay until the company returned to profitability.

In 1989 EMC accelerated the transition from a supplier of memory enhancement products to a provider of storage solutions. In addition to offering the only plug-compatible alternative to IBM’s DASD (Direct Access Storage Device) system for its mini-computers, EMC entered the IBM mainframe storage market with the introduction of a mainframe compatible solid state disk subsystem, the Orion. In June 1989, the company entered into a $100 million agreement with StorageTek to be the exclusive distributor of a customized version of Orion which featured the industry’s first 4-megabyte RAM chip technology. StorageTek, a major vendor of IBM-compatible tape storage systems, was attempting to develop an innovative disk storage system for mainframes and saw the Orion as a valuable intermediate product. Continuing problems from 1988, added to a large inventory write-down related to discontinued lines, resulted in an $18 million loss for 1989, but the new product focus led to strong revenue growth during the year, reaching $40 million in the fourth quarter, and set the stage for explosive growth in the 1990’s. Also of major importance to the future of EMC was its adoption in 1989 of a Continuous Improvement program, driven by Michael C. Ruettgers, the newly-appointed President and COO, aimed at eliminating waste, cutting costs and improving quality throughout the organization. Every employee attended a three-month Conway Quality Management seminar.

Mainframe Disk Storage Systems

In the late 1980’s, the IBM mainframe computer disk storage industry consisted of IBM and its principal mainframe competitors, Amdahl, Hitachi and Fujitsu, all of which manufactured and sold "plug-compatible" disk storage systems. Historically, IBM controlled the mainframe storage marketplace through its proprietary disk drive technology, which offered the most advanced performance in such features as speed and areal density. The critical element of IBM’s control was the tight linkage between the physical characteristics of the disk drive and the image that the mainframe operating system wanted to see - a one-to-one physical mapping. IBM’s competitors were forced to match those physical characteristics in order to achieve compatibility, i.e. be plug-compatible. Its control of the technology allowed IBM to hold an 80% market share through-out the 1980’s, a position considered to be virtually un-assailable.

EMC took a different approach to challenge IBM’s hegemony, enabled by the rapidly evolving technology and reduced costs of commodity PC components - SCSI, microprocessors, DRAM chips and disk drives. By 1990, microprocessor speed and power had increased ten times since 1985, and the cost of DRAM was plummeting. Additionally, the capacity, performance and reliability of small, 5.25-inch commodity disk drives, originally designed for use on personal computers and workstations, was increasing. By linking groups of inexpensive 5.25 drives together, in Redundant Arrays of Independent (or Inexpensive) Disks (RAID), EMC could provide sufficient capacity and reliability to substitute for the plug-compatible mainframe storage that used 14" proprietary disks.

The concept of using arrays of commodity disk drives to create a storage system evolved from EMC’s solid state storage product, Orion, which used a SCSI-attached disk storage to ensure fault tolerance. By adding multiple disks onto the back end, it transformed the solid state technology into a large write/read cache. By putting arrays of multiple parallel processors between the cache memory and the disks and between the cache memory and the mainframe host computer, EMC created a novel storage architecture that offered substantial gains in system speed as well as other potential features not yet appreciated in the mainframe marketplace.

In contrast to the efforts of plug-compatible manufacturers to duplicate IBM’s one-to-one physical mapping, EMC created a "logical to physical mapping," using the controller, cache and algorithms. EMC argued that the storage system didn’t need to look like an IBM disk drive system, it was only necessary that the computing processor "think" that it looked like one, i.e., receive data in the same format.

The cache principle had been in use for many years on host processor systems where cache functioned as a high-speed intermediary between CPU’s and main memory to store frequently-used data. Historically in storage systems, however, data was always written directly to the disks and was transferred to a small cache when it was read. The concept of writing data to cache was relatively recent in 1990. The availability of powerful microprocessors enabled EMC to add intelligence relatively cheaply and to develop algorithms for finding data in cache and moving ("destaging") data from cache to disks with extreme efficiency, resulting in significant improvement in overall system performance.

In the fall of 1990, EMC introduced the Symmetrix 4200 Integrated Cached Disk Array (ICDA) system, 24 gigabyte RAID storage with 256 megabytes of cache and a 32-processor controller. The large cache and the destaging algorithms in the controllers enabled the Symmetrix to attain system speeds that were significantly faster than competitors’ products, yet took up only twenty percent of the floor space. The availability of RAID 1 disk mirroring, battery back-up to make cache non-volatile, "write/verify" capability to check data, and constant disk defect monitoring all contributed to high reliability.

With the continued fast pace of advances in the technology of disk drives, SCSI interfaces, and microprocessors, EMC rapidly increased the capacity, speed and reliability of Symmetrix releases. The product’s reputation for performance - it ran some applications orders of magnitude faster - was also growing quickly, and the value of performance as a key criterion of storage came to be better appreciated. In 1994, the announcement of the Symmetrix 5500 allowed EMC to move from its initial niche marketing to a mainstream position, and the rapid pace of product enhancements through-out the early and mid-1990’s drove dramatic gains in market share. From five percent of mainframe terabytes shipped in 1992, the Symmetrix had nearly a third of the market by 1994 and became the leader in 1995 with 41 percent. By 1993, the company had discontinued the Orion, and its memory and cache upgrade products, to focus exclusively on disk storage.

Storage Software

Beginning in 1994, EMC enhanced the functionality of its Symmetrix systems with intelligent software programs that added special data management, protection and business continuance features. The first storage software product was the innovative Symmetrix Remote Data Facility (SRDF) which provided fast disaster recovery and remote data mirroring. As software increasingly served as basis for differentiation in a market of growing competition and hardware commoditization, EMC focused on developing this area as a key source of revenue. In 1997, more than seventy percent of the company’s engineers were dedicated to software development. Software sales rose from $20 million in 1995 to $445 million in 1998, making EMC the fastest growing major software company in the industry.

Enterprise Storage

In November 1995, EMC announced the Symmetrix 3000 to simultaneously support mainframe and heterogeneous open systems data (UNIX and NT) on the same system. At that time, most corporations had data on a variety of computing platforms - mainframes, large UNIX servers, and smaller Windows NT. Symmetrix Enterprise Storage enabled heterogeneous platforms to utilize the same storage device.

A significant issue in the corporate IT world in the late 1990’s was the tension between centralization and de-centralization of computing power, data storage and IT systems management. The countervailing trends were enabled by technology and driven by organizational and business forces, and the evolving direction had significant implications for EMC and its vision of enterprise storage.

Starting in the early 1970’s, IT followed a path of decentralization as computing power and its associated data storage migrated from the data center mainframe, to the departmental server and to the workstation and personal computer. Business needs initially drove the trend, and budget structures sustained it. Mini-computers and workstations enabled departmental managers and individuals to control their own projects and accomplish time-sensitive business tasks more efficiently than centralized IT departments, who often responded slowly to internal customer requirements. The cost of purchasing a few mini-computers was less than that of installing or upgrading a mainframe system. Moreover, in most companies the cost of a large centralized system required high level approval, while the purchase of a few workstations or a new server could usually be done without external review. The spread of decentralized computing brought a proliferation of decentralized storage systems attached to departmental servers, rendering large quantities of corporate data inaccessible to the majority of users.

The move to re-centralize corporate IT in the mid-1990’s arose from the realization that distributed computing had become unmanageable. A substantial amount of business-critical computing was running on departmental Unix and NT servers and was not subject to uniform data protection and security protocols. Multiple client-server systems with banks of PC’s encouraged the proliferation of varying types and releases of software and different versions of shared files. The true cost of ownership of de-centralized systems was substantially greater than forecast, with total costs often calculated to be three to ten times the cost of acquisition. Additionally, as new applications started to capture the value of corporate data, companies realized the potentially huge opportunity costs of having information scattered throughout an organization.

Driven by the convergence of technology, location and function, the formerly distinct mainframe and mid- to high-end open system (Unix and NT) segments were rapidly merging into a single heterogeneous enterprise segment. RAID was the common technology for all these systems, which increasingly were being co-located in data centers and running mission-critical applications. Non-enterprise segments included distributed Unix, NT and NOS server-attached storage, and internal disk storage on low-end servers and desktops. For 1998, EMC held approximately 50% of the mainframe market, 30% of the heterogeneous enterprise market and 11.5% of the total market, which included internal disk storage.

 


 






Legal information © 2000 EMC Corporation. All rights reserved. About this site. Feedback.