What Is a Mainframe?

Before there were servers or personal computers, there were mainframes. Nicknamed the ‘big iron,’ a mainframe is a high-performance computing system that’s used to rapidly process large amounts of data in real time.

Mainframes have been around since the 1950s, back when they took up entire rooms, so it’s easy to mistake them for outmoded pieces of technology. But they have only gotten more efficient. Of the roughly 10,000 mainframes in use today, more than two-thirds of them power Fortune 100 companies.

 

What Is a Mainframe?

A mainframe is a high-performance computing system designed to rapidly process large amounts of data. Known for their high throughput and reliability, these refrigerator-sized machines are built with specialized hardware that enable large-scale batch computing with efficient input-output connectivity, allowing them to process up to one trillion online calculations per day.

Mainframes often serve as the central hub for critical business applications — like processing credit card transactions, airline reservations and inventory tracking — that support a range of industries, from finance and healthcare to the government sector. According to IBM, the primary manufacturer of mainframe hardware, 92 of the world’s top 100 banks rely on mainframes, which are also responsible for processing nearly all credit card transactions.

“Mainframes continue to be the technology backbone of the world’s economy,” Petra Goude, senior vice president of core enterprise and zCloud at mainframe modernization company Kyndryl, told Built In. Despite its age, the mainframe “remains the right platform for mission-critical workloads,” Goude said.

Mainframes earned their name from the large cabinets, or main frames, in which they were originally stored. Back then, these much larger, earlier iterations weighed several tons and were used to solve scientific calculations and for military applications. As tech evolved over the years, many companies transitioned away from mainframes in favor of distributed computing alternatives and cloud-based networks. Even so, mainframes remain a staple in IBM’s portfolio, as well as those of other major players like BMC, Precisely, Compuware and CA Technologies.

Traditionally, mainframes were a sort of “black box” that isolated data from other infrastructure systems — resulting in compatibility issues, Jacky Hofbauer, the president and chief security officer at mainframe-management software company Zetaly, said. However, novel tools that permit the access, collection and sharing of mainframe data using APIs are emerging, allowing hybrid-cloud modernization to breathe new life into an old relic. 

Related ReadingWhat Is Exascale Computing?

 

How Are Mainframes Used?

Mainframes commonly serve as central fixtures to large business operations, thanks to their ability to reliably process massive volumes of data in real time and host mission-critical applications. Even with the advent of cloud computing, which has contributed to the reduction in mainframes that exist today, mainframes are still responsible for 70 percent of the world’s IT production workloads.

“You’ll find them at the heart of operations where reliability can’t be compromised,” said Darian Shimy, a former head of engineering at Square and Weebly and founder of school-fundraising platform FutureFund. Whether it’s in banking, healthcare or government services, “their job is to keep the digital aspects of heavy-duty industries running smoothly.”

Here’s how mainframes are currently applied across several industries:

Financial Services

Banks and financial institutions use mainframes to process transactions, manage accounts and execute trades. If you’ve ever made an ATM withdrawal or swiped a credit card, you’ve interacted with an in-network card-reading device that communicates information, such as account balances and transaction records, to and from a mainframe.

Healthcare

Since health records have become digitized, mainframes serve as the central hub of HIPPA-compliant patient management systems and information exchange. While they provide secure storage of diagnostics and medical imaging, they also assist doctors in transferring encrypted patient records back and forth to ensure continuity of care.

Government

Governments rely on mainframes to host mission-critical applications and manage large volumes of data. These machines keep track of citizen records, process tax filings, administer social welfare programs and support national security.

Retail

Point-of-sale systems, inventory tracking and customer relationship management platforms use mainframes to keep retail operations in sync. The same systems that digitally calculate totals and authorize payments at the register for a seamless shopping experience can also maintain real-time records of products as they move through the supply chain and keep track of loyalty program points using a database of customer profiles.

Travel

Airlines use mainframes to handle millions of reservation requests daily. On passenger-facing interfaces, they deliver real-time updates to flight schedules, seat assignments and pricing information, then process boarding passes, seat upgrades and any additional baggage handling requests at check in. For airport staff, these centralized platforms handle everything from flight planning and crew scheduling to aircraft maintenance.

Manufacturing

Mainframes support a number of manufacturing processes. Similar to retail, these systems are equipped to manage inventory and supply chain logistics as well production planning, quality assurance and distribution. No matter the production line, a mainframe enables seamless data exchange and decision-making across a company.

Education

In educational institutions, mainframes serve as a centralized platform that manages student records, administrative tasks, courses and online learning platforms. They are used to tally enrollment and grades as well as data analysis or simulation modeling in research labs and master organizers in library management.

Related ReadingWhat Is a Data Center? What Goes on Inside One?

 

Mainframe Features

Mainframes are known for the following standout features.

High Performance

Mainframes are designed to deliver exceptional performance, capable of processing large volumes of data and supporting intensive computing workloads. They feature powerful processors, large memory capacities and high-speed input-output (I/O) subsystems to ensure fast and efficient data processing.

Reliability

A mainframe’s reliability delivers continuous operation at minimal downtime. Built with redundant components and fault-tolerant designs, mainframes have advanced error detection and verification mechanisms that prevent system failures, ensuring uninterrupted service and a near-guarantee to anytime database access.

Scalability

Mainframes can be built out to accommodate growing computing needs and increasing workloads. This is done by scaling vertically, where additional processors, memory and storage capacity are added; or scaling horizontally, where multiple mainframe systems are connected in a parallel configuration to increase processing power and capacity.

“Not only are [mainframes] capable of bearing considerable workloads,” Hofbauer said, “they are also highly scalable and can increase capacity without experiencing performance degradation.”

Security 

Mainframes are “the most securable platforms” in computing, Goude said. Their built-in components feature only what can be described as “encryption everywhere” authentication mechanisms, which keeps data encrypted not only on storage devices but also in the processor and as it traverses the network. Viruses are also virtually unknown on mainframes.

Centralization

Mainframes unify an organization’s data silos altogether in one place. Using a centralized approach, administrators can monitor, configure and manage system resources from a single point of control.

 

How Do Mainframes Work?

Modern mainframe systems process small data in large batches. Instead of working on one massive complex problem at a time like a supercomputer, mainframes handle high volumes of simple computations.

They do this with the help of specialized computer processing units (CPUs). These hardware components are the “brain” of the computer that houses the electrical circuitry necessary to interpret and process inputs, store memory and carry out instructions in order to run programs and deliver outputs.

Multiple CPUs are often housed together in a central processing complex, and work in tandem with additional processors called system assistance processors (SAPs), tasked with moving data through I/O channels between the mainframe and external, in-network devices. The I/O then computes an output based on the requested information, and reports it to the original source.

For example, when a user inserts their card into an ATM card reader, the device retrieves that user’s debit account information from a bank’s mainframe over a network. It simultaneously validates the user’s identity, checks the availability of funds, processes the withdrawal amount and updates the accounts accordingly.

Modern mainframe systems use redundant circuitry in order to achieve high fault tolerance and availability without a single point of failure, according to Craig Wilson, a global technology officer at mission-critical support service provider Top Gun Technology.

“If a fault is detected,” he added, “the operation is retried or recovered using error correction routines.” It can also move the data to a twin power circuit to complete the computation.

Wilson said that an entry-level mainframe may have a single CPU node, known as “a drawer,” with a single I/O node, whereas higher-end models may feature multiple CPU and I/O nodes that trade off data and processing workloads. Take IBM’s latest iteration of its popular Z series, the IBM z16, for example, which features four CPUs and 12 I/O drawers for a multiframe model, and four CPUs with 3 I/O drawers for the single-frame standard model.

“This allows businesses of all sizes to realize the benefits of using mainframe computers,” Wilson added.

Related ReadingWhat Is Grid Computing?

 

Examples of Mainframes

IBM Z series

IBM’s latest series introduces AI inferencing to the mainframe space. The z16 model features an IBM Telum microprocessor at its core and industry-first, on-chip integrated accelerators with generative AI that predicts and automates tasks, delivering unprecedented speed and scale at low latency. It can support up to 40 terabytes of memory, process 25 billion transactions per day and can even survive an 8.0-magnitude earthquake.

FUJITSU Server GS21

FUJITSU launched its GS21 series in 2018, featuring a super-scale and medium-to-large version of the server. The super-scale model can host up to 16 CPUs with 256 gigabytes of storage and a maximum of 256 channels. So while it’s a 20-percent upgrade in performance compared to previous iterations, the hardware required to run such a machine has reduced by 40 percent. The Japan-based company plans to discontinue mainframe sales by 2030, but will continue providing parts and maintenance support through 2035.

UNIVAC 9000 series

Now discontinued, the UNIVAC 9000s are a relic of the 1960s, when mainframes first came onto the scene. The Sperry Rand series featured four generations of machines that were built with small scale integrated circuits using DTL technology and plate-wire memory. They were sold as a “flexible tape and disk oriented computer system featuring multi programming, real time capabilities and versatile possibilities for data telecommunication.” As featured in Technikum29’s Living Museum, one 9400 model — originally used in a Cologne-based Industrial complex computer center, then repurposed and given to a school the following decade — is still operational today.

Frequently Asked Questions