Content Addressable File Store
Hardware disk storage with built-in search capability.
The Content Addressable File Store (CAFS) was a disk storage system from International Computers Limited (ICL) that could search data directly on the drive. It was built to solve a bottleneck: disks could feed data to a processor far faster than that processor could sift through it for matching records. Work on CAFS began in the late 1960s at ICL’s Research and Advanced Development Centre, led by Gordon Scarrott. This followed a field study by George Coulouris and John Evans at Imperial College and Queen Mary College, which showed that adding search logic to a disk controller could greatly speed up large database tasks.
In its earliest form, the search circuitry sat inside the disk head. A standalone CAFS unit was used by a handful of customers during the 1970s, including BT Directory Enquiries. The device was later turned into a standard product, and by 1982 it was a built-in feature of ICL’s 2900 series and Series 39 mainframes. To cut costs and exploit faster hardware, the search logic had moved to the disk controller by then. A query written in a high-level language could be compiled into a search specification and sent directly to the controller. This first worked with ICL’s own Querymaster language and the IDMS database, and later with the ICL VME version of the Ingres relational database. In 1985, ICL received the Queen’s Award for Technological Achievement for CAFS.
Adoption was limited because CAFS required a specific data layout on disk and had to know that layout. Integrating it with database products often meant changing page layouts, which was expensive, especially as the market shifted toward third-party database software. Managing data integrity in a multi-user environment was also tricky, since a CAFS search ran without awareness of locks or caches maintained by the database.
In 1991, ICL released a version of CAFS for its DRS 6000 line, called SCAFS (Son of CAFS). Unlike the mainframe version, this ran as custom firmware on a standard microprocessor. Software supporting third-party databases such as Ingres, Informix, and Oracle was sold as the Ingres Search Accelerator. Each third-party product needed modification and came with a dummy SCAFS interface library that would be swapped for the real ICL product. The technology was also licensed to IBM for use with DB2 on the RS/6000.
- Developer
- International Computers Limited (ICL)
- Development started
- late 1960s
- Key researchers
- Gordon Scarrott, George Coulouris, John Evans
- Initial form
- search logic built into the disk head
- First standalone installation
- 1970s, including BT Directory Enquiries
- Incorporated into mainframes
- 1982, within ICL 2900 series and Series 39
- Queen's award for technological achievem
- 1985
Lore & Background
The Content Addressable File Store (CAFS) originated from research by George Coulouris and John Evans at Imperial College and Queen Mary College, whose field study on database systems and applications indicated that including search logic in the disk controller could yield substantial performance gains in large-scale database applications. Development began in ICL's Research and Advanced Development Centre under Gordon Scarrott in the late 1960s. In its initial form, the search logic was built into the disk head, and a standalone CAFS device was installed with a few customers, including BT Directory Enquiries, during the 1970s. The device was subsequently productised and in 1982 incorporated as a standard feature within ICL's 2900 series and Series 39 mainframes. By this stage, to reduce costs and take advantage of increased hardware speeds, the search logic was moved into the disk controller. A query expressed in a high-level query language could be compiled into a search specification sent to the disk controller for execution. Initially this capability was integrated into ICL's own Querymaster query language, which worked with the IDMS database; later it was integrated into the ICL VME port of the Ingres relational database. ICL received the Queen's Award for Technological Achievement for CAFS in 1985.
Reader's Guide
One factor limiting CAFS adoption was that the device needed to know the layout of data on disk and placed constraints on that layout. Integrating database products with CAFS often involved changing page layout, making integration very expensive, especially as the market trended toward third-party database software. Managing data integrity in a concurrent environment also required close attention, since a CAFS search would execute without knowledge of locks and caches maintained by the database software. In 1991, ICL launched a version for its DRS 6000 family, known as SCAFS (Son of CAFS), implemented using custom firmware on an industry-standard microprocessor. Software supporting third-party databases including Ingres, Informix and Oracle was marketed as the Ingres Search Accelerator. Each third-party product required modification and was supplied with a dummy SCAFS interface library to be replaced by the ICL product. The technology was also licensed to IBM for use with DB2 on the RS/6000. The device eventually became obsolete as processor speeds increased, removing the original justification that a central processor could not search data as fast as the disk subsystem could deliver it. Larger memory sizes also meant many medium-sized databases could be kept entirely in memory, removing any mass market for SCAFS and making it uneconomic.
Did You Know?
- CAFS development began in the late 1960s under Gordon Scarrott at ICL's Research and Advanced Development Centre.
- The initial form of CAFS had search logic built into the disk head.
- A standalone CAFS device was installed with BT Directory Enquiries during the 1970s.
Frequently Asked Questions
What is the Content Addressable File Store?
CAFS was a disk storage system produced by International Computers Limited that embedded search logic directly into the drive hardware, allowing matching records to be located on the disk rather than pulled all the way to the processor for scanning.
Who created CAFS?
The project was led by Gordon Scarrott at ICL's Research and Advanced Development Centre beginning in the late 1960s, building on earlier field research conducted by George Coulouris and John Evans at Imperial College and Queen Mary College.
What problem was CAFS designed to solve?
Disks could deliver data to a processor far faster than that processor could sift through it for matching records, creating a serious throughput bottleneck; CAFS moved the matching logic into the disk controller to eliminate that choke point.
When and where was CAFS first installed in the wild?
The first standalone deployment took place during the 1970s, with British Telecom's Directory Enquiries service standing out as one of the earliest real-world applications of the technology.
How did CAFS transition into mainframe systems?
By 1982 the technology had been folded into ICL's 2900 series and Series 39 mainframes, shifting from a standalone disk unit to an integrated component of larger computer systems.
More in Computer Storage 1-24
Spotted an error? Know more?
Reader corrections go straight into our review queue. Suggest an edit · How this site is sourced