The history of the operating system

In the history of the operating system, I comment on the history of the relatively recent calculator from the viewpoint of operating system (the OS).

I provide function group necessary to control the hardware of function group and the computer that most application programs of the computer need the OS and use, and to synchronize it. An early computer did not have the OS and completely grasped the hardware specifications of the computer even if the individual program executed a normal task, and it was necessary to control the peripheral devices such as a printer or the punched card leader by paying own expenses. As hardware and an application program became complicated, the OS became required.

Background

The early computer was a mainframe and did not have the OS. The user of this time used a machine alone and I prepared for a program and data in form of the punched card and magnetic tape person hole tape beforehand and came over. I let a machine read the program and let you execute a program until it is completed or crashes. The debugging of the program used toggle switch and the lamp of the control panel. It is said that the alane Turing was good at such a work in one of the early computers Manchester Mark I. He has already arrived at the basic concept of "the operating system" from a principle of the all-around Turing machine.

The programmer developed symbolic language, an assembler, a compiler and I wrote it with the program cord that a program written in a binary directly was encoded till then and came to convert it into a binary automatically. Furthermore, I gather up a supporting program handling input and output as a library and store it to a punched card and a magnetic tape and I link with a user's program and come to use it. The structure is a beginning of modern operating system. However, the machine was able to yet carry out only one job at a time at that point. Drew tape to the wash-line pole in British Cambridge University and performed job management of the human power to attach a colored clothespin to each, and to show priority.

When the performance of the computer improved, time to take grew big to really succeed the use from a user to a user than the time when I calculated. I checked the clock which hung over the wall and charged it, but a computer came to take the log automatically till then. The orchid cue was a cue of the users who lined up before a door literally, but it was a reservation list, and they came to pile up the punched card of plural jobs to a leader, and the OS came to finally choose order online, to do a magnetic tape unit automatically. A program developer operated a machine directly and carried out a job, but the exclusive duty operator who performed the maintenance of the machine came to replace it and it was hand-operated and came to rarely set a task once. When the calculation center came to be satisfied as business, it became the problem that data were lost in bugs of the input and output, and the device vendor came to cannot but strengthen a run time library to prevent misuse of the system resource. The automatic monitoring came to amount to the pages that made prints, the number that I punched of cards, the number that I read of cards, the disk space that I used as well as the use of CPU situation, and the function that the intervention of the operator including job practice magnetic tape exchange told about a necessary phenomenon was born. For security capability, an inspection evidence function to record which program accessed which file was increased by the OS and came to prevent unjust access such as a technology calculation program accessing a salary account book file.

All these functions became a part of the complete operating system. And the run time libraries evolved and came to start before job injection and accepted a job and controlled the practice and recorded the use of resource situation and came to have the function to assign it again, and to carry out the next job of the resource when a job completed it. Called it with a monitor or a monitor program, and the program to be stationed in this background, and to manage the whole system came to be called the operating system later.

It may be said that a program to perform management of the hardware, scheduling of the software, resource monitoring is the ancestor of the user intention OS of the later personal computer times. However, the thing which the OS means changes with the times. An old car had neither the speedometer nor the radio nor the air-conditioner, but they became supplied later. Like it, an optional software function became the standard function of many OS packages. But the application such as database management system and the spreadsheet is still optional and is sold separately. Therefore, speaking of the OS, I include a graphical user interface in the present age and comprise utility group and come to point to the system including application and setting tools such as text editor or the File Manager.

The true descendant of the early OS is a part called "a kernel" now. In the technical field, the OS in the meaning from old times is still kept. This is because development of the embedded operating system of the electronic equipment with various data handling from a portable device to an industrial robot and a real-time control system is still performed lively, and most do not carry out the application of the front end which a user sees. The modern embedded OS is not a thing far apart in comparison with the ancestors of the OS of the 1950s that much either.

Mainframe era

It is GM-NAA I/O that the study section [¹] of General Motors developed for IBM 704 in 1956 to be said to be the world's first OS put to practical use [²]. A customer developed the OS for many other early IBM mainframes by oneself [³].

The early operating system was very various, and a vendor and a customer brought about various OS's for a specific mainframe. Even if it was the same vendor, a command system and an assembly operation, facility for the debugging made plural totally different OS's. Generally, whenever a maker developed a new model, the new operating system was developed and I revised application with human power each time and compiled it again and evaluated it again.

OS of the computer made in IBM

Such state continued until the 1960s, but IBM which distinguished itself as a hardware vendor stopped the development of various systems which I developed till then and came to concentrate on development of the System/360 series. It is all the series that standardized the input and output architecture using the same instruction set from a small machine to a large-scale machine. IBM was going to develop single operating system OS/360 for this series. The problem that occurred by development of OS/360 half becomes the legend and is described in "the myth in a person month" of Frederick Brooks that it may be said with the classic of the software engineering. In OS/360, it was of the OS family without covering the whole series by the really single OS because performance of the whole series was too different, and software development delayed [⁴]; [⁵].

IBM will release two following operating system for a long time.

The OS for scale out of OS/360 - and large scales. Three kinds of constitution options existed.
- A product for PCP - initial users. There is not a resource for the multiprogramming.
- Scale system use out of MFT -. It became MFT-II in OS/360 Release 15/16. Furthermore, succession included OS/VS1, but I develop it in the 1980s and am completed.
- MVT - large scale system use. I am similar in PCP and MFT and most of (the transplant of the program allows it between 3 people without compiling it again), but memory management is refined and comprises time sharing facility TSO. Succession of MVT includes z/OS.
In a product for small models of DOS/360 - System/360, there are some succession, and there is z/VSE. It is greatly different from OS/360.

Because IBM always maintained compatibility, even the program that I developed in the 60s is executable without a change in current z/VSE (program case for DOS/360) and z/OS (in the case of a program for MFT and MVT).

IBM developed time sharing system TSS/360 for System/360 Model 67, but did not release it formally.

The OS that a third party developed for System/360 and System/370 includes Michigan Terminal System (MTS) or MUSIC/SP.

OS of other mainframes

The Control Data corporation (CDC) develops the SCOPE operating system for the batch processing in the 1960s and develops the MACE operating system for the time sharing afterwards. MACE became the base of later KRONOS. I developed KRONOS and the NOS operating system of the succession in cooperation with the University of Minnesota in the 1970s. The NOS supported batch processing and time sharing at the same time. Like many commercial time sharing systems, the interface expanded the interface of DTSS which was the time sharing system of the dawn.

In the late 1970s, CDC developed PLATO system which adopted plasma display and a long-distance time-share ring network in cooperation with University of Illinois. PLATO was extremely innovative for those days, and the application such as real-time chat and the graphical game of the multi user type was enabled by the shared memory model of the programming language TUTOR.

UNIVAC which produced commercial computers from an infantility developed a series of OS's called EXEC. Like the OS for early other mainframes, I manage the magnetic drum memory, magnetic disk, card reader, line printer by the batch processing-oriented OS. When it was the 1970s, UNIVAC developed Real-Time Basic (RTB) system for large-scale time sharing. This imitated DTSS and the Dartmouth BASIC system, too.

In B5000 which Burroughs announced in 1961, I adopted the OS called MCP (Master Control Program). B5000 is written without using a machine language and the assembly language with the stack machine which was designed to support a high-level language even in a core part of the OS. MCP uses ESPOL which is a dialect of ALGOL by the world's first OS that the whole was written in a high-level language. But ESPOL included a special sentence corresponding to the instruction set of B5000. Besides, there is advanced characteristic in MCP [^{NB 1}] and implements virtual storage as the commercial OS quickly. The OS that rewrote MCP for B5000 is still used in the ClearPath/MCP series of Unisys.

GE developed the OS called GECOS (General Electric Comprehensive Operating Supervisor) for the GE-600 series. When Honeywell got a computer section of GE, I changed my name with GCOS (General Comprehensive Operating System).

Project MAC of Massachusetts Institute of Technology developed Multics in cooperation with GE and Bell Laboratories. The new features such as ring protections were introduced in Multics.

Digital Equipment Corporation (DEC) developed large number of OS's for various computer series. For example, there is a time sharing system called TOPS-10 and TOPS-20 for PDP-10 of 36 bits. TOPS-10 was used well at a university before Unix spread, and even the early ARPANET was used well.

When performance of the hardware improved, I operated similar software by plural systems, and it became often said that I transplanted software in 異機種間 from the late 1960s through late 1970s. Therefore I used microprogramming by the early system and even if it was the architecture that became the base, it was the same from software and might show it. For example, in System/360 of IBM, the model after 360/40 becomes the microprogram implementation (except 360/165 and 360/168). However, a method to secure compatibility of the application by other means soon became mainstream.

Sudden rise of minicomputer and Unix

Development of Unix began in Bell Laboratories of AT&T at the about the end of 1960s. The early printing block is available easily free and will spread widely because the correction was easy. It became the procurement requirements in the subsidiary group of AT&T again. If even the new architecture supported even C language because it was written by C language, the transplant was easy. It was adopted with the second generation minicomputer and the first generation work station widely by this transplant-related good point. Because it was adopted widely, I will illustrate a way of thinking to sit on various hardware platforms, and to use the same operating system conceptually. The current UNIX was proprietary software, and only a company and the organization which could pay license charges used it, but, as for Unix, it was one in the origin of the open source exercise.

In addition, DEC developed some operating system for PDP-11 of 16 bits. It is simple RT-11, RSTS of the time sharing type, RSX-11 family of the Real-time Operating System. I developed VMS for the VAX of 32 bits.

The system which developed for the same period includes Pick which Microdata Corporation developed. Originally Pick begins as a supporting program of the database applications and is the example which evolved to the OS.

An 8 bits home computer and game console

Home computer

Commodore 64, Apple II, 8 bits plane of Atari, Amstrad CPC, most of 8 bits home computers of the 1980s such as ZX Spectrum loaded the OS such as CP/M and GEOS from a floppy disk and were able to use it, but generally often used it without the OS. In many cases, BASIC interpreter was sold to ROM in the state that it stored, and it functioned as the simple OS. Simple file operation (deletion, copy), the format of the disk, a road and practice of the application were possible.

I was often used in entertainment and an education purpose, and it was rare to use those machines for business and a technology calculation, and there was few that the true OS was required.

Because another reason why the OS was not required was a machine of the single user who executed a single task, and the primary storage used the processor of low 8 bits of the performance small (generally there were many cases from 64KB to 128KB of at 256KB from 4KB), this was because it could not ignore an extra overhead to suffer from by using the OS.

The application such as word processors at the time controlled the whole hardware by paying own expenses, and it was similar with the video game.

A game console and game

(the の clone and the derivative were analog "pop"), this flattered it with complete digital electronic equipment, and almost all game consoles and arcade game consoles which designed it after 1980, and were produced had only minimum BIOS to John, SEGA master system, SNK neo-geo-といった product. In addition, Nintendo Entertainment System and the clone gain success by a design not to have BIOS at all.

PCエンジンを初めとする現代的なゲーム機では、最小限のBIOSを備えるのが一般的となり、メモリーカード管理、CD再生、コピーガードといった機能を備え、時には開発者向けのライブラリを備えるようになった。しかし、これらを真のOSと呼ぶことはほとんどない。

特筆すべき例外としてドリームキャストがある。プレイステーションと同様に最小限のBIOSを備えているが、ゲームディスクから Windows CE オペレーティングシステムをロードでき、PC/AT互換機からのゲーム移植を容易にしていた。Xboxでは、Microsoft Windows に修正を加えたOSがバックグラウンドで密かに動作しており、PCの一種と言えなくもない。ドリームキャストやその後のゲーム機で動作する Linux もある。

ソニーは初代プレイステーション向けのソフトウェア開発キット「ネットやろうぜ!」をリリースしている。通常のPC上で動作する一連の開発ツールと、PCと接続してプログラムをダウンロードして実行できる黒いプレイステーションが含まれていた。これには両プラットフォーム上で動作するOSを必要とする。

1990年代までのゲーム機やアーケードゲームはBIOSのような機能を組み込んでいたが、プレイステーション以降はそれがどんどん洗練されていき、OSと呼んでも差し支えないものになっていった。

パーソナルコンピュータ時代: Apple、Amiga、PC/MS/DR-DOSとその後

マイクロプロセッサが開発されたことで中小企業や趣味で使える低価格のコンピュータが登場し、共通規格化されたハードウェア部品が広く使われるようになった（S100バス、SS-50、Apple II、ISA、PCIといったバス）。すると、それらを制御する「標準」のオペレーティングシステムの必要性が増していった。初期の重要なOSとして、デジタルリサーチが開発した8080/8085/Z80といったCPU向けのCP/M-80がある。DECのいくつかのOS（主にPDP-11用OS）をベースとした設計となっている。マイクロソフトの最初のOSであるM-DOSもPDP-11のOSに倣い、マイクロプロセッサ向けに開発された。IBMのMS-DOSまたはPC DOSは、当初CP/M-80をベースとしていた。このころのマシンには小さなブートプログラムを格納したROMがあり、それがディスクからOS自体をロードする。IBM PCのBIOSはこの考え方を拡張したもので、1981年に最初のIBM PCが登場してから20年間でより多くの機能が追加されていった。

表示装置とプロセッサのコストが低減されてくると、多くのOSでグラフィカルなユーザインタフェース（GUI）を実用的なコストで提供できるようになっていった。Unix向けには汎用的な X Window System が登場し、他にも Microsoft Windows、ラジオシャックの Color Computer 向けの OS-9 Level II/MultiVue、コモドールの AmigaOS、アップルの Mac OS、IBMのOS/2といったグラフィカルなインタフェースを備えたOSが登場した。GUI搭載オペレーティング環境のインスピレーションとなったのは、ゼロックスのパロアルト研究所で1970年代に開発された暫定Dynabookと言われ、それを多くのベンダーが真似ていった。

暫定Dynabook（Smalltalk）の生み出したもう一つの遺産であるオブジェクト指向については、ネクスト・コンピュータのNEXTSTEPが引き継ぎ､発展させた｡

仮想化の発展

元々、OSはハードウェア上で直接動作し、アプリケーションへのサービスを提供していた。IBM System/360 Model 67 上の CP-67と、System/370上の Virtual Machine Facility/370 (VM/370) が仮想機械の考え方を導入し、OSがハードウェアを直接制御するのではなく、ハイパーバイザの制御下でOSを実行する方式を採用した。このテクノロジーをパーソナルコンピュータ上で広めたのがVMwareである。その後、仮想機械とモニターとOSの境界は徐々にぼやけていった。

ハイパーバイザーはより複雑化していき、自前のAPI^[6]、メモリ管理やファイルシステム^[7]を持つに至っている。
仮想化はOSの重要な機能の1つとなっている。例えば、Microsoft Windows Server 2008 の Hyper-V、HP-UX の HP Integrity Virtual Machines などがある。
IBMのPOWER5やPOWER6ベースのサーバなどのシステムでは、ハイパーバイザーがオプションではなく必須となっている^[8]。
VMモニター上で直接動作するよう再設計されたアプリケーションもある^[9]。

最近ではかつてOSが占めていた位置を仮想化ソフトウェアが占めるようになってきており、ハードウェアリソース（プロセッサ、メモリ、周辺機器）の管理機能、スケジューリング機能、システム管理者のためのインタフェースなどを提供している。

脚注

^ B5000はフェランティのAtlasと同時代のマシンである。

出典

^ See Rand Corporation publication by Robert Patrick (PDF)
^ "Timeline of Computer History: 1956: Software". コンピュータ歴史博物館. 2008年5月25日閲覧。
^ A Brief History of Linux
^ Johnston (April 1, 2005). "VSE: A Look at the Past 40 Years". z/Journal (Thomas Communications, Inc.) (April/May 2005).
^ Chuck Boyer, The 360 Revolution (PDF)
^ "VMware API". VMware. 2008年11月26日閲覧。
^ "VMware file system". 2008年11月26日閲覧。
^ "PowerVM Virtualization on IBM System p: Introduction and Configuration". 2008年11月26日閲覧。
^ "JRockit's Liquid VM could be the first real Java OS". 2008年11月26日閲覧。

参考文献

Neal Stephenson (November 1999). In the Beginning... Was the Command Line. Harper Perennial. ISBN 0380815931. OSの歴史を解説した読み物

TranPedia

Wikipedia Translation from Japanese to English

2017년 2월 13일 월요일