We use them daily but we are not aware of them. They help the science evolve and better, that new discoveries are possible. Without them, the world today would be very different. The Mainframe they are a category of computers unknown to many, but essential in modern computing.
Today we delve into the world of the mainframe, taking advantage of the 50th anniversary of its creation by IBM. Equipment which, like everyone else’s technology, have evolved very remarkable, moving from those old valves vacuum to the new transistors, have cabinets with control panels can be managed from anywhere in the world.
Going back to the prehistory of computing
Early 20th century computers were calculators, used to accelerate operations professionals
Before entering into matter with the mainframe, it is necessary to put some context to this topic. Early 20th century computers were mere calculators that the researchers of the time used to speed up operations. Machines with an electronic Basic, far removed from that we know today, but they ran and helped then studies. We can tell which served to make mathematics accounts pass the paper to the computer, speeding them up and, in many cases, ensuring the results to human errors.
IBM 601, calculator based on card micro-perforated (1931, via Wikipedia)
Obviously human aspirations were increasingly higher. If we have a computer capable of running 1,000 operations per second, the next step to perform is immediate: search one exceeds him in performance.
In the 1930s, IBM already manufactured computers intended for large entities, usually public administration (Governments, armies or universities), which reserved several rooms in their buildings for these computers. Input and output methods were microperforated business cards, too far away from the keyboard, mouse, or display current. Information is encrypted and is recorded on the cards through small physical holes, which were that the computer detected and was able to read through a code. After having made certain operations, the computer went to punch another card which showed the results.
IBM 801 Bank Proof, machine to keep track of cheques (IBM, 1934)
These teams were relatively small, of a few square meters of surface. On these lines you have the IBM 801 Bank Proof, a system which IBM sought to automate some work related to banking: check bank checks, approve them and take an economic record in the set. A curious machine that just arrived a few years after the crash of 29, specifically in 1934, and was one of the culprits that IBM had a rate of manufacture of between 5 to 10 million microperforated business cards a day.
But was not allowed to be toys for what was going to come sometime later. The first use of IBM to the term ‘mainframe’ is in 1944, When the North American company and well known Harvard University developed the ASCC, Automatic Sequence Controlled Calculator, also known as The Harvard Mark I, and it began to take shape in 1939.
Harvard Mark I
One of the first ‘big’ teams from IBM, spent 15 years operating in Harvard admitting up to 23 decimal numbers.
The Mark I It was a team of great dimensions that occupied a huge room inside Harvard, and can be described as a huge synchronous calculator that could work in parallel, using algorithms that could perform four basic arithmetic operations.
The Harvard Mark I, or ASCC, launched in 1944, is considered the first mainframe market
According to IBM, Mark I operated with numbers up to 23 decimal using the 60 physical records for constants and 72 results temporary counters, which were physically built by 765.000 components and about 800 kilometers of cable. And Yes, of course the Mark I received as input microperforated business cards, although you could also add information through paper or mechanical switches; Moreover, output was also performed with microperforated business cards or automated, directly to paper, through mechanical typewriters.
The Mark I started its work on August 7, 1944, after six months of mounting facilities of Harvard and several delays caused by World War II, and was in operation for fifteen years.
The Mark I is considered the first mainframe of the story, and after him came many more. While Frank Sinatra and Elvis Presley began to sound in the American radios, IBM engineers continued with the idea of the Mark I and raised the challenge to start its evolution and improve it.
It was then when she was born the SSEC, Selective Sequence Electronic Calculator, a system with mechanical parts (vacuum valves) and electronic (relays), whose design was started just after the Mark I, becoming operational in January 1948. Its mission was not far from which was carried out with the Mark I, being also a great calculator which first use was the calculation of the positions of the moon with respect to the planets of the Solar System. This task performed it during the first six months, and after that it began to be rented to different companies for multiple projects.
Note the dimensions of the equipment on the sides, and in the middle of the room control panels. You can find the history of this and other images at the Columbia University website.
The Mark I was born during the war could be a coincidence, if it is true that the war world has always been closely linked to the evolution of technology. Many projects are great secrets and, unfortunately, no we never see them, but there are others that have been confirmed.
For example in full war of Korea (1950-1953), IBM introduced the IBM 701 in 1952, a team whose development was agreed upon with the US Government in order to be used for the design of aerospace ships, ammunition and also for nuclear development.
One of the IBM 701 of the Lawrence Livermore National Laboratory, in 1954. Once again repeats the distribution of cabinets and central control (image via Flickr) table.
‘Typical’ installation of an IBM 701 (image via trailing-edge)
In addition, IBM 701 is a before and an after in the history to be considered as the first digital information processing system. While in earlier computers was the user that along with the information added operations to perform on the 701 programs were stored in an internal memory, as it is done in modern computing. Also he is considered one of the key pieces for the abandonment of the microperforated business cards, in addition to the seed that initiates a new era of modern computing.
IBM 701 ushers in a new era, being the first digital information processing system. Was presented in 1952.
According to the information of the company, the 701 is 25 times faster than the SSEC while it occupies a quarter of your space. Only sold 19 units, available between 1952 and 1955, belonging to the following customers:
|Machine number||Shipped to||Give you||Note|
|1||IBM World Headquarters, New York, N.Y..||Dec. 20, 1952|
|2||University of California., Los Alamos, N.M.||Mar. 23, 1953||Used for hydrodynamics calculations.|
|3||Lockheed Aircraft Company, Glendale, Cal.||Apr. 24, 1953||Mathematics Analysis Department used the 701 for problems in aircraft design, such as aerodynamic performance and stability, thermal dynamics and structural and flight dynamics. Production data on 701s handled by Lockheed completo Factory Data Processing Group included project base schedule preparation, parts scheduling, shop order writing, direct labor hour forecasting and parts activity ledgers.|
|4||National Security Agency, Washington, D.C..||Apr. 28, 1953|
|5||Douglas Aircraft Company, Santa Monica, Cal.||May 20, 1953||Arrived on May 23, 1953, aboard a DC-6A aircraft. The 701 was used to get the DC-7 into production months ahead of schedule. It solved engineering and scientific problems on all Douglas commercial aircraft, including the DC-6B, DC-7, DC-7 c and the development of DC-8.|
|6||General Electric Company., Lockland, Ohio||May 27, 1953|
|7||Convair, Fort Worth, Tex.||Jul. 22, 1953|
|8||U.S. Navy, Inyokern, lime.||Aug. 27, 1953||Used to calculate rocket and missile performance and to simulate flight conditions of these devices at the U.S. Naval Ordnance Test Station China Lake.|
|9||United Aircraft, East Hartford, Conn..||Sep. 18, 1953|
|10||North American Aviation, Santa Monica, Cal.||Oct. 9, 1953||Handled engineering problems, from basic configuration selection through aerodynamic and structural design to the analysis of flight test data.|
|11||RAND Corporation, Santa Monica, Cal.||Oct. 30, 1953||Used to solve wide variety of problems in economics, mathematics, aircraft, missiles, electronics, nuclear energy and social sciences. Later moved to West Los Angeles.|
|12||Boeing Corporation, Seattle, Wash.||Nov. 20, 1953||Used to assist engineers and designers in solving problems in aerodynamics, and structural development, flight testing and stress of supersonic and jet aircraft and guided missiles.|
|13||University of California, Los Alamos, N.M.||Dec. 19, 1953|
|14||Douglas Aircraft Company, The Segundo, Cal.||Jan. 8, 1954||Engineering problems solved on U.S. Navy A3D Skywarrior, A4D Skyhawk and F4D Skyray programs, and USAF C-133 and RB – 66 programs.|
|15||Naval Aviation Supply, Philadelphia, Pa.||Feb. 19, 1954|
|16||University of California, Livermore, Cal.||Apr. 9, 1954|
|17||General Motors Corporation, Detroit, Mich..||Apr. 23, 1954|
|18||Lockheed Aircraft Company, Glendale, Cal.||Jun. 30, 1954||Mathematics Analysis Department used the 701 for problems in aircraft design, such as aerodynamic performance and stability, thermal dynamics and structural and flight dynamics. Production data on 701s handled by Lockheed completo Factory Data Processing Group included project base schedule preparation, parts scheduling, shop order writing, direct labor hour forecasting and parts activity ledgers.|
|19||U.S. Weather Bureau, Washington, D.C..||Feb. 28, 1955||Produced from spare parts.|
Arrive the transistors and the mainframe continues to grow
Since the 1950s, mainframes have continued evolving with the help of design engineers have been permanently looking for the way to improve them and adapt them to the new challenges and inventions of science.
Part of these developments were looking to improve the physical part and certain components in order to achieve greater efficiency and to minimize their physical error rate. This line of thinking was that in 1956 he develop the first magnetic hard drive, a monster of several kilograms of weight that leased for $3,200 per month, and that was installed for the first time on the IBM RAMAC 305.
Yes, this cylinder you see there is the first hard drive on the market (image via ExtremeTech)
The first mechanical hard disks, and then transistors: mainframes changed radically in the 1960s.
Few years later executed one of the most important innovations in use in current technology. You know that the term ‘bug’ it comes from its definition in English, whose translation is ‘bug’, ‘bug’ or similar individuals in nature. This is because in those old computers machinery got insects that altering the proper functioning of the system.
Avoid those ‘ bugs’, or at least that they cease to exist as they were known at the time, was one of the greatest challenges in the 60’s to leave behind the technology of ” vacuum valves, indispensable for the computers of the time. It is when they arrive transistors and starts the ‘digital age’.
Mainframe Manager opening this was the IBM 7000, an evolution on the 700 that changed the vacuum valves transistors, and which came to the market in 1964.
The seed was already buried in the ground and was a matter of giving away it and give it some time. The IBM 7000 Series had some success in the market, but above all were the basis on which later, in 1964, they built the IBM System/360, It would significantly change the way to reach customers.
Arrive 60 and becomes the color, customization and the different product ranges. The result? The S/360 became a hole in many companies (image via 360info).
Instead there are few teams made almost by hand, the S/360 were designed in such a way that any multiple configurations to adjust the budgets that could have different customers, as well as also to the requirements of all of them and even color, being possible to customize them to suit the image of the different companies.
The IBM System/360 wasn’t just brilliant all-in-one computer mainframe; It was stylish too. It was available in five standard colors, although the ASB Bank data center in New Zealand took advantage of ordering it in a custom yellow.
Mainframes began arriving to companies of all types, and not only to entities of public court with large budgets, as until then. Need a wardrobe? Maybe two? Those who need, or those who can afford.
But the success of the System/360 was not only based on the ability to scale them to what the customer needs. IBM also directed them to serve as multipurpose machines. While previous generations were designed as supercalculadoras, from the S/360 mainframes began to be machines that were used for much more than calculate numbers.
The transition along with the PC and the advent of the Internet
IBM engineers kept the idea of the System/360 with new generations in the successive years. In 1970 came the System/370 for those who thought that it would be a good idea that people who already have a S/360 could continue running your software on a new S/370. They were retrocompatibles, based on the same principles but evolved and adapted to modern times. As a curious detail, the operating system of the S/360 and S/370 already based architecture with 32-bit registers, which in the domestic arena even continue using currently.
The control panel, at the bottom, and data entry, to the right. Behind the photographer would have more machines (image via Forbes).
Also given a major coincidence that later seized from IBM. The System/370 coincided with the birth of the PC, equipment much easier but also more economical. As a reference, the S/370-158 and S/370-168, both in 1972, had a cost of 200,000 and 400,000 dollars (of the time) respectively, while the Apple I in 1975 was introduced by the famous 666,66 dollars.
The S/370 remained on the market for two decades
The System/370 were one of the more long-lived families of the history of computing, staying on the market for two decades. IBM, aware that had a good product, focused on improving on both the hardware (new processors, adaptation to new interfaces) as in the software (operating systems specifically designed for IBM, introduction of the first Linux made ‘bespoke’). They also began to emerge the clones IBM equipment, defined as computers designed and built by other manufacturers (Hitachi, Fujitsu, Siemens and Mitsubishi were perhaps the best-known) but the same software provided by IBM-compatible.
These ‘clone’ remained on the market during the 1980s, and began to expand throughout the domestic environment (who does not remember the 80286 manufactured by AMD?), remained for the next generation.
With the Cowboys and the cola sold everywhere, in the 90’s, IBM released the System/390. Same structure of name which of course brought some new features (passed using bipolar transistors instead of CMOS, for example), but who maintained the essence of the S/370 and S/360 previous: remained retrocompatibles software (Yes, in 1999 could be run on a S/390 a program developed for a S/360 in 1964 without having to adapt the code) and used the same 32-bit CISC architecture. Also there were also multiple families, configurations and ranges, a constant which remains to the present day.
An IBM S/390, already “only closet” (image via Corestone)
A significant difference is that, while the S/360 and S/370 machines with an own control panel, the S/390 started to adopt a different format. Cabinets of, generally, large dimensions, where all communication was done from an external client. The PC, with its rapid growth and the threat posed to mainframes, motivated this change that made these teams need to have a good channel for the exchange of information.
The turn of the century and the name change
Although the IBM design team had in mind always the evolution of many of the components, the basis on which successive generations were sustained was the CISC 32-bit architecture which, as we have seen, was maintained between the S/360, S/370 and S/390 for a period of 40 years.
The arrival of the new century brought major changes to IBM. The S/390 were happened by the System z, a new name that also included a new architecture, the z/Architecture.
Installing a zEnterprise and its expansion, BladeCenter Extension, in 2010. What is said four summers ago.
The z/Architecture, also developed by IBM, was introduced in 2000. They gave the jump to 64-bit and, once again, they kept the backward compatibility with previous generations, including with the S/360 and assuming 40 years of compatible software.
Software of 60 in a 20th century zEnterprise? Retrocompatibles for more than 40 years
Also introduced a new limit in RAM (up to 2 exabytes) fundamental professional teams of this caliber, and of course the software, which could now access records of greater capacity, greater efficiency in the processing of information affecting.
Something that began to glimpse on the S/390 ended up exploiting the System z: the parallelization It became a fundamental pillar of this new generation. To be competitive and taking into account the experiences of the previous years, these new machines are designed in such a way that they support settings which in some cases they exceed 100 processors per Cabinet, each one with four, six or more cores, and above the terabyte of RAM memory.
Currently, in addition to the System z, IBM also distributes mainframes with AMD Opteron chips (System x), Intel Xeon (System x, BladeCenter) or POWER (Power Sustems, Power Blade), its own architecture inherited from the PowerPC of yesteryear. There are dozens of different models, which in turn are widely customizable and adaptable to the requirements of each customer, both technological and economic.
The mainframe: are still there but you can’t see them
The mainframe exist and follow the order of the day, and in fact we use them constantly. We do not see them, but they are there. If you stop to think, these lines which you are probably reading are stored on a server somewhere, and that server will consist of several mainframes which provide connection 24/7 of this and many other websites.
Great services have their own datacenters (Google, Apple), while others outsource it. For example, Amazon is one of the best positioned in this with its Amazon Web Services, providing servers and cache systems essential for the Web today.
Mainframes have evolved systems belonging to public entities (universities, Governments, armies) to be the base hardware of Internet and communications. The world of smartphones has made us even more connected, not only in our homes but also in (almost) anywhere in the world. Smartphones must be added (increasingly common in cities) sensor networks, vehicles (cars increasingly incorporate more technology) or electrical appliances that are beginning to be connected (see this Berg).
There is still a long way to go, but the trend is that practically all end connected to a server, where there will be a mainframe.
IBM zEnterprise zEC12
Another zEnterprise, 2012, alongside its BladeCenter Extension zEC12. Special attention to the design of the equipment, authentic beauty geek.
Equally, we must also mention the importance of the mainframes for the world of science. This cutting equipment is used to perform highly complex mathematical simulations, weather forecasts or analysis of information that would be impossible to meet without these super-fast computers.
We have been using them for years, without almost realizing.
Mainframes are several years being transparent ‘Center’ of our connections. We do not see them, but they are there. Its features will not change in the coming years, although Yes they begin to see some modifications important that they can affect pillars of these teams.
One of them is the energy consumption, a very valued parameter in recent years due to increases in the cost of energy in the world. Customers are looking for teams tighter consumer that allows them to reduce their electricity bill, and therefore manufacturers are faced with the challenge of lowering the energy requirements of your equipment at the same time that maintaining good performance. It is important to mention that the power consumption depends not only on the team as such, if not that for example the cooling of the room is also essential (supercomputers tend to settle in completely refrigerated rooms to improve heat dissipation), which often assumed much of the total bill.
The world of mainframes and servers in general, in particular, is in the process of change, with ARM as the next great implementation.
One of the most important changes that are arising in the mainframe market is to evolve from the classic POWER architectures, x 86 or z that we mentioned earlier, to ARM, that it has in its favour several factors. Consume a fraction of the energy of its competitors, emits heat that often can dissipate passively (no active fans) and they are cheap. Obviously have the counterpart that his performance is much lower that the of the above alternatives, something that is fighting by installing multiple ARM chips and running them simultaneously, in parallel.
The HP Moonshot, racks that can be configured with ARM processors