Disk pack
A disk pack is a removable magnetic disk storage medium developed for early mainframe computers, consisting of multiple rigid platters stacked vertically within a sealed, portable cartridge that could be easily inserted into and removed from compatible disk drives.[1] Introduced by IBM in October 1962 with the Model 1311 disk drive, it marked a significant advancement over fixed-disk systems by enabling data portability, offline archiving, and enhanced security through physical removal of media.[1] Each pack typically featured six 14-inch diameter platters made of aluminum coated with magnetic oxide, providing an initial storage capacity of about 2 million characters (roughly 2 megabytes) across 10 recording surfaces, while weighing approximately 9.4 pounds for easy handling.[1][2] The design of disk packs revolutionized data management in the 1960s and 1970s by reducing reliance on bulky, non-removable drives like the earlier IBM RAMAC, which used 24-inch disks fixed in place.[1] IBM's innovation, credited to engineers including Lou Stevens and Jack Harker, shrank the platter size from 24 inches to 14 inches, making storage more compact and cost-effective for mid-sized systems.[1] Subsequent models, such as the 1964 IBM 2311 and the 1970 IBM 3330, increased capacities to 7.25 megabytes and 100–200 megabytes per pack, respectively, incorporating advancements like voice coil actuators for faster access times.[2] The packs were encased in durable polycarbonate shells to protect against dust and damage, with read/write heads in the drive latching onto the platters only during operation to minimize wear.[1] Adoption extended beyond IBM, as the standardized 14-inch form factor spurred a competitive industry of plug-compatible manufacturers (PCMs) like Memorex, Telex, and Century Data by the late 1960s, alongside media suppliers such as 3M and BASF.[1] Vendors including DEC (with the RP04 drive) and ICL (EDS-200 packs for the 2900 series) integrated disk packs into their mainframes, making them a staple for business and scientific computing until the early 1980s.[1][2] Their impact included fostering the growth of the hard disk drive market, but by the mid-1990s, they were largely supplanted by non-removable, higher-density Winchester drives and later technologies like optical and solid-state storage.[1] Today, disk packs represent a foundational milestone in the evolution of digital data storage, preserved in computer history collections for their role in enabling scalable, user-friendly mass storage.[2]History and Development
Origins in Early Magnetic Disk Storage
The development of magnetic disk storage began in earnest at IBM in the early 1950s, culminating in the announcement of the IBM 305 RAMAC on September 14, 1956, the world's first commercial computer system featuring a hard disk drive.[3] The associated IBM 350 Disk Storage Unit consisted of 50 fixed, 24-inch diameter platters coated with magnetic oxide, providing a total capacity of approximately 5 million characters—equivalent to about 5 MB in modern terms—and enabling random access to data at speeds far surpassing previous sequential storage methods like magnetic tape or drums.[4] These non-removable disks were housed in a massive unit weighing over one ton, designed primarily for business applications such as accounting and inventory control.[5] Despite its groundbreaking random-access capabilities, the fixed-disk design of the RAMAC presented significant challenges that limited its practicality. The system's enormous size and weight made it immovable once installed, severely restricting portability and complicating data sharing or backup across different machines, as physical relocation of the entire unit was infeasible.[3] High manufacturing and operational costs—estimated at around $10,000 per megabyte in 1956 dollars—further hindered widespread adoption, while the fixed configuration offered limited scalability, as expanding storage required entirely new, expensive installations rather than incremental additions.[6] These limitations spurred early experiments at IBM's San Jose Laboratory, established in 1952 specifically for magnetic disk research, where engineers explored modular disk assemblies to improve flexibility. In 1953, proposals included arrays of continuously rotating disks on straight shafts with independent access stations, alongside innovations like air-bearing read-write heads that achieved 51 bits per inch by mid-year.[7] The first successful data transfer occurred on February 10, 1954, using a test bed, though challenges such as disk wobble at high speeds and maintaining precise head-to-disk spacing persisted.[7] These foundational efforts in the 1950s laid the groundwork for future advancements, including the shift toward removable storage solutions in the following decade.Introduction of Removable Disk Packs
The introduction of removable disk packs marked a pivotal advancement in magnetic disk storage, building on the fixed-disk concepts pioneered by the IBM RAMAC system in 1956. In October 1962, IBM announced the 1311 Disk Storage Drive, the first commercial product to feature interchangeable media in the form of the 1316 Disk Pack.[1] This pack consisted of six 14-inch diameter disks, providing a total capacity of approximately 2.8 million characters (equivalent to 2.8 MB), and weighed just 10 pounds, enabling easy portability between compatible drives.[8] Designed for medium-scale business and scientific computers, the 1311 drive allowed users to quickly swap packs, facilitating offline data management and transport without the need to dismantle entire storage units.[9] The removable disk pack concept was formalized through a U.S. patent issued in 1965 to IBM engineer Robert E. Pattison, titled "Portable Memory for Data Processing Machine" (U.S. Patent No. 3,176,281).[10] This patent highlighted key benefits such as unlimited offline storage capacity through multiple interchangeable packs and enhanced data security via physical removal and transport, protecting sensitive information from unauthorized access during system maintenance or relocation. By allowing data to be stored externally and exchanged as needed, the design addressed limitations of fixed-disk systems, where media was permanently integrated and vulnerable to on-site risks. Initial adoption of the 1311 and 1316 Disk Pack occurred primarily in mainframe environments, where it enabled efficient data sharing across multiple systems by simply exchanging packs between drives.[9] This portability reduced operational downtime associated with fixed drives, as organizations could prepare backup or working datasets offline and insert them rapidly, minimizing interruptions in data processing workflows.[11] The technology's immediate market impact was evident in its integration with IBM's early 1960s computing lineup, establishing removable media as a standard for scalable storage in enterprise settings and influencing subsequent disk drive innovations.Evolution and Adoption in Computing Systems
The introduction of removable disk packs with the IBM 1311 in 1962 laid the groundwork for portable storage in computing systems.[1] In 1964, IBM integrated disk packs into its System/360 family through the 2311 drives, which utilized the same 1316 packs as the 1311 but increased storage capacity to 7.25 MB per pack through improvements in recording density.[12][9] This enhancement supported the System/360's modular architecture, enabling scalable storage for diverse applications and driving widespread enterprise adoption by standardizing compatible peripherals across IBM's mainframe lineup.[13][14] By the late 1960s, disk packs saw growing usage in batch processing for high-volume transaction handling, database management systems like the Integrated Data Store for structured data access, and scientific computing workloads on System/360-compatible platforms.[15][16][17] Their removable design facilitated data sharing and backup, making them a standard for offline storage in both mainframes and emerging minicomputers during the 1970s.[18][19] The market for disk packs expanded beyond IBM in the late 1960s and 1970s, with vendors like Control Data Corporation (CDC) producing compatible systems and packs for IBM 2314 and 3330 drives, as well as proprietary formats that gained traction in peripheral markets.[20][21] This interoperability fueled industry-wide proliferation, with disk pack drives from multiple manufacturers dominating computer rooms and supporting peak adoption around 1970 as storage demands surged in enterprise and research environments.[2][22]Design and Components
Physical Structure and Materials
A disk pack typically comprised a stack of 6 to 11 rigid aluminum platters, each with a standard 14-inch diameter, precisely finished and coated on their recording surfaces with a thin layer of magnetic iron oxide particles suspended in a plastic binder to enable data storage.[23][9] These platters were assembled vertically on a central aluminum alloy spindle hub, which facilitated precise alignment and rotation when mounted on a compatible drive, with spacers between disks to promote airflow and maintain separation.[24] The entire assembly rotated at 2,400 RPM during operation, providing the necessary speed for data access in systems like the IBM 2311.[9] The platters and hub were enclosed in a sealed, portable canister made of durable polycarbonate plastic or lightweight metal, featuring a locking lid to safeguard the components during handling and storage.[25][9] Internal environmental seals around the enclosure prevented dust and contaminant ingress, critical for maintaining the integrity of the delicate magnetic surfaces and minimizing the risk of head crashes.[26] Protective features, such as raised edges or bumpers on the platter rims, served as crash stops to guard against damage if the heads inadvertently contacted the surfaces during non-operational states.[7] Overall dimensions of a typical pack, such as the IBM 1316 model, measured about 4 inches in height for the disk stack within a canister roughly 14 inches tall, with a weight of approximately 9 to 10 pounds to allow safe manual transport.[1][9] Integrated carry handles on the canister exterior enabled operators to easily move the pack between storage and drives, emphasizing its design for portability in data processing environments.[1] Early models like the 1316 provided around 2 MB of storage capacity, underscoring the pack's role in compact, removable media.[7]Magnetic Media and Data Encoding
Disk packs utilized aluminum platters coated on both sides with a thin layer of ferromagnetic iron oxide particles embedded in a binder, enabling magnetic storage of data through the alignment of these particles in response to write head fields. This oxide coating, typically gamma ferric oxide, provided the necessary magnetic properties for retaining binary data as localized magnetic domains representing 0s and 1s. Both surfaces of each platter were coated to maximize storage capacity, with the media designed for durability during repeated mounting and unmounting in removable configurations.[27][28] Data organization on the disk pack surfaces followed the Count Key Data (CKD) format, in which concentric tracks contain an index marker followed by one or more variable-length records, each preceded by count and optional key fields, enabling flexible record addressing in mainframe environments. In the CKD format, each track begins with a home address (index) and contains records structured as count (location and length), optional key (index), and data fields, supporting variable record sizes typical of early mainframe applications.[29][30] This structure simplified data management in early computing systems, enabling the disk pack to serve as interchangeable media across compatible drives. Early disk packs employed Frequency Modulation (FM) encoding. Later models, such as the IBM 3330, used Modified Frequency Modulation (MFM) as an advancement over FM, doubling the effective data rate by more efficiently encoding clock and data bits without requiring separate clock tracks. In MFM, transitions in the magnetic flux represented data bits, achieving linear recording densities of approximately 4,040 bits per inch (bpi) on models like the IBM 3330, contributing to areal densities around 775,000 bits per square inch when combined with track densities of 192 tracks per inch. This encoding was particularly suited to the oxide media, balancing signal reliability with storage efficiency.[31][32] To ensure data integrity, especially in removable environments prone to handling errors, each sector incorporated error detection via cyclic redundancy check (CRC) or error-correcting codes (ECC), such as those replacing CRC in the IBM 3330 for single-bit error correction. Additionally, wider inter-track spacing, often including guard bands, was implemented to minimize adjacent track interference from stray magnetic fields or slight misalignments during pack mounting, thereby reducing crosstalk and read errors without significantly sacrificing capacity.[33][9]Integration with Drive Mechanisms
Disk packs interface mechanically with host disk drives through a precise latching mechanism that secures the pack onto the drive's central spindle, ensuring alignment and stability during operation. In the IBM 2311 Disk Storage Drive, for instance, the spindle-lock mechanism engages to hold the disk pack assembly firmly in place, utilizing clamp rings to prevent horizontal shifting of the disks relative to the hub.[29] This design allows the pack's hub to align vertically and horizontally with the spindle, which rotates the disks at a constant speed, such as 2,400 RPM in the 2311 model.[29] While early systems relied on mechanical latching, later variants incorporated more automated securing processes to enhance reliability and ease of interchangeability across compatible drives. Electrically, disk packs integrate indirectly through the drive's connections to a storage control unit (SCU), employing multi-pin connectors for power, data signals, and control lines that adhere to IBM's proprietary bus standards for System/360 and subsequent architectures. These interfaces include power supplies such as +36V DC for the drive motor and hydraulics in models like the 2311, alongside signal lines on a file bus (e.g., lines 0-7 for cylinder and head addressing) and control lines for operations like module select and seek initiation.[29] Once mounted, the drive's actuator arms position the read/write heads over the pack's tracks using integrated servo systems for precise access. Early drives like the 2311 employed a hydraulic actuator with access and extend pistons to move the head carriage across up to 203 cylinders at variable speeds (e.g., 7.9 inches/second for intermediate positioning).[29] In contrast, the IBM 3330 series advanced to an electromagnetic voice-coil actuator with closed-loop track-following servo control, utilizing a dedicated servo surface on the disk pack to provide real-time position and timing information, enabling higher density and accuracy independent of environmental factors like temperature.[34][31] This servo mechanism ensures the heads maintain alignment with tracks as the pack spins, supporting reliable data access in removable media environments.Operation and Usage
Mounting and Unloading Procedures
The mounting procedure for a disk pack begins with powering down the associated disk storage drive to ensure safety and prevent mechanical interference. Operators must first verify that the drive is stopped by toggling the start/stop switch to the off position and confirming the ready indicator is extinguished.[35] Next, the access cover or module drawer is opened, allowing inspection of the spindle area for any obstructions. Visual checks for damage on the disk pack, such as dents, scratches, or contamination on the recording surfaces, are essential before proceeding; damaged packs should not be mounted and require inspection by qualified personnel.[36][35] With the pack's protective bottom cover removed by turning its locking knob counterclockwise, the operator aligns the pack's keyway with the drive's spindle pin and gently lowers it onto the spindle until fully seated, ensuring even contact to avoid misalignment or surface damage.[35] The top cover is then rotated clockwise in the "ON" direction until firm resistance is felt, securing the pack in place via its latching mechanism. The module drawer or cover is closed firmly to maintain a dust seal, and the start/stop switch is set to the start position.[36] The drive initializes by spinning up the pack, typically reaching operational speed in about 60 seconds, at which point the ready indicator illuminates, signaling completion.[35] Throughout the process, operations occur in a clean, dust-free environment to minimize contamination risks to the magnetic surfaces.[36] Unloading follows a reverse sequence to protect the pack and drive components. The start/stop switch is toggled to stop, halting the spindle motor and allowing the pack to decelerate fully, which requires waiting until the ready indicator turns off—typically 15 to 60 seconds depending on the model.[36][35] The access cover or drawer is then opened, and any brush assembly extending into the pack area is gently moved aside if present. The top cover is placed over the pack and rotated counterclockwise in the "OFF" direction for at least two full turns to release the latch.[36] The pack is lifted straight upward by the top cover handle, avoiding contact with recording surfaces, and the bottom cover is securely fastened to reassemble the unit for storage.[35] The cover or drawer is closed if no immediate remounting is planned, and the pack is stored in a designated clean cabinet or shelf at controlled temperatures (60°F to 90°F) and humidity (10% to 80%) to prevent environmental degradation.[36] Operators handle packs solely by edges or designated grips to evade static discharge and particulate transfer, with any suspected contamination necessitating professional cleaning.[35] These protocols, exemplified in systems like the IBM 2311 and 2314 drives using 1316 or 2316 packs, emphasize precision to safeguard data integrity in data center operations.[36][35]Alignment and Calibration Processes
Alignment and calibration processes for disk packs were essential to ensure precise head positioning relative to data tracks, enabling reliable data access across interchangeable packs and drives. These procedures were typically executed by trained customer engineers following the mounting of a new or replacement disk pack, compensating for mechanical variances in head assembly, spindle alignment, or pack manufacturing tolerances. The primary goal was to center the read/write heads on reference tracks, minimizing off-track errors that could lead to read/write failures or data corruption.[37] Alignment relied on specialized tools and a dedicated CE (Customer Engineer) disk pack, which contained pre-recorded test tracks for verification without risking data on user packs. The CE pack, identifiable by its yellow top cover, featured eccentric tracks—such as those at cylinders 72 and 74, offset by ±0.008 inches—to assess head centering. Procedures began with connecting an oscilloscope (e.g., Tektronix 561 or 453) to monitor read amplifier outputs at specific test points, typically showing peak-to-peak signals of 300 mV (±40 mV) for cylinder transducers. Engineers used a head adjusting Bristol wrench (P/N 2108626) to loosen clamp screws on the head arms, preset adjustment screws to 0.188 inches, and iteratively rotate the backstop screw while observing the oscilloscope waveform for symmetric, equal-amplitude lobes indicating centered positioning. For tracking adjustments, cylinder 73 served as the master alignment track, with read data pulses expected at +3 to +4 Vdc and 100 ns width. Heads were cleaned with 91% isopropyl alcohol and lint-free tissues prior to adjustment to remove contaminants affecting signal quality.[37][9] Calibration routines incorporated reference signals from the CE pack's index and cylinder transducers to fine-tune positioning, often involving a recalibrate command that homed the access mechanism to cylinder 000 for detent seating. Tunnel erase features on the heads, which narrowed written tracks from 0.008 inches to 0.005 inches, provided inherent tolerance for minor misalignments by reducing adjacent-track interference. In high-use environments, periodic realignment was recommended every few months or after 10,000 hours of operation to account for wear. Common issues included off-track errors manifesting as asymmetric waveforms (lobe ratios >4:1) or low head output (<51 mV average), often due to arm shifts during clamping or head-to-disk contact indicated by audible "tinging" or dark stains on media. Solutions entailed iterative screw adjustments for balance, head replacement if output remained low, and grounding checks (<10 ohms) to eliminate noise; spare tracks on cylinders 201 and 203 allowed error recovery without full recalibration.[29][37]Data Access and Performance Characteristics
Data access in disk packs involved a combination of seek time, rotational latency, and data transfer phases, which collectively determined the overall performance during read and write operations. Typical seek times ranged from 50 to 200 milliseconds across early models, with the IBM 1311 exhibiting an average of 250 ms for random access over 100 cylinders, while later designs like the IBM 2311 achieved an average of 75 ms and the IBM 2314 reduced this to 60 ms in select configurations.[11][29][38] Rotational latency averaged 12.5 ms at the common 2,400 RPM speed of models such as the IBM 2311 and 2314, representing half the time for one full disk revolution of 25 ms.[29][38] Transfer rates varied by model and encoding density but generally fell in the range of 300-800 KB/s, with the IBM 2311 delivering 156 KB/s and the IBM 2314 doubling this to 312 KB/s through improved recording techniques.[39][38] These rates prioritized sequential access patterns, as random reads or writes incurred significant overhead from repeated seeks and latencies, making disk packs more efficient for linear data processing than fragmented workloads.[39] Reliability during operation was constrained by mechanical vulnerabilities inherent to removable designs, with mean time between failures (MTBF) typically around 1,000-5,000 hours in early disk pack systems.[40] Head crashes, where read/write heads contacted the spinning platters, posed a major risk in removable setups due to potential misalignment or contamination from pack handling, often leading to data loss across multiple tracks. Additionally, disk packs were susceptible to environmental shocks during active use, such as vibrations or jolts that could cause heads to skip tracks or crash, exacerbating failure rates in non-sealed environments.Variants and Legacy
Key IBM Models and Specifications
The IBM 1311 Disk Storage Drive, introduced in 1962, marked the debut of removable disk pack technology for mainframe systems, utilizing the 1316 disk pack with six 14-inch platters providing 2 million characters of storage capacity (approximately 2 MB in modern terms), or up to 2.98 million characters (approximately 2.2 MB) with the optional Track Record feature.[41] It operated at 1,500 RPM with a data transfer rate of 89,000 bytes per second, enabling random access for applications on systems like the IBM 1401.[1] The IBM 2311 Disk Storage Drive, announced in 1964 for the System/360 family, improved upon the 1311 by employing the same 1316 disk pack but with enhanced electronics, supporting six recording platters for a capacity of 7.25 million bytes.[12] Operating at 2,400 RPM, it achieved a transfer rate of 156,000 bytes per second, roughly 75% faster than its predecessor, while maintaining compatibility with existing packs to facilitate customer upgrades.[42] This model supported up to eight drives per control unit, scaling total storage to 58 MB.[39] The IBM 3330 Series, launched in 1970, represented a significant advancement with the 3336 disk pack featuring 11 thinner platters to increase areal density, available in Model 1 (100 MB capacity) and Model 11 (200 MB capacity) variants.[43] Running at 3,600 RPM, these models delivered a transfer rate of 806,000 bytes per second and incorporated track-following servo technology for precise head positioning, reducing average access time to 30 ms.[44] The series supported up to 16 drives per subsystem, providing up to 1.6 GB of online storage, and its interface design paved the way for the Storage Module Device (SMD) standard, enabling third-party compatibility.[20]| Model | Introduction Year | Platters (Recording) | Capacity per Pack | RPM | Transfer Rate (bytes/s) |
|---|---|---|---|---|---|
| 1311 | 1962 | 6 | 2 MB | 1,500 | 89,000 |
| 2311 | 1964 | 6 | 7.25 MB | 2,400 | 156,000 |
| 3330 (Model 1) | 1970 | 11 | 100 MB | 3,600 | 806,000 |
| 3330 (Model 11) | 1970 | 11 | 200 MB | 3,600 | 806,000 |