When Cloud Is Better Than Mainframe (and When It’s Not)
Broadcom Mainframe Software CTO George DiCandio stresses that hybrid IT success depends on picking the right environment for the right workload
As one of the fastest, most secure and most reliable computing platforms ever, the mainframe is awesome. But it’s not well-suited for every task. For good reason, cloud and the accessible application development models that run on it have become the de facto choice for many new apps, use cases and small-to-midsize companies.
For example, a customer-facing mobile app that needs frequent updates and rapid iteration belongs in the cloud. A high-volume payment processing system that requires absolute consistency and auditability still belongs on the mainframe.
Still, when large enterprises and governments need to process enormous volumes of sensitive data in milliseconds, with maximum security and integrity, 70% of Fortune 500 companies still lean on mainframes to get the job done. That’s why “hybrid cloud” has become an enterprise buzzword in recent years, combining a “best of both worlds” approach to successful IT strategies.
But when is cloud better than mainframe, and when is mainframe better than cloud? Which apps and workloads belong on which platform? And what’s the most cost-effective platform for any given use case? Let’s break it down.
Better on Cloud
When people think of fast development, public cloud is what usually comes to mind. This is the biggest reason for its meteoric rise over the last 20 years. The ability to develop relatively functional, internet-facing applications over a weekend is one of its perceived strengths.
This is possible because of how easy it is to build an application out of standard components like databases, message queues and web servers. Many of these components are open-source, freely documented and available as hosted services that can be quickly configured on any number of cloud platforms.
Speed, Agility and Initial Affordability
The cloud is easier to deploy, iterate on and scale for many modern applications. The former deploys faster, can be quickly changed and isolates individual services from each other, which makes test automation much more straightforward and frictionless.
The cloud enjoys lower acquisition costs than the mainframe. With cloud applications, programmers can start small, inexpensively (in many cases free), and even scale relatively seamlessly by clustering nodes dynamically as the application needs additional compute power.
These three dimensions of speed, agility and initial affordability make the cloud a very attractive and powerful platform for developers and application owners, especially for consumer-facing web, mobile, and internet applications. Even from the perspective of advocating for mainframes in the right setting, there is absolutely no better way to run these front-end applications. In fact, the only reason to run a web application on a mainframe would be for additional high-performance or business continuity capabilities, but this is usually not the case.
The Challenge of Cost Comparison
Lastly, cloud computing is often cited as being much less expensive than mainframe computing. In terms of enterprise-grade security that reliably runs a high volume of consistent transactions, however, this is simply not the case.
Although raw computing is typically more expensive on the mainframe, cloud does not have the same built-in services as the mainframe, namely backup, encryption, transactional integrity and disaster recovery. Matching this level of service requires additional infrastructure and expenses, without the same proven track record.
Additionally, cloud software is often open source and free but unsupported. Although paid-for, mainframe software comes with enterprise support, which must be factored into the total cost of any platform (as should on-premises costs such as electricity). The mainframe also scales better and becomes cheaper over large transactional applications that serve millions of people, as opposed to thousands.
Because of these factors, it is difficult to compare the true costs of the mainframe versus the cloud. What is clear is that cloud offers faster and more cost-effective entry for smaller, consumer-facing applications.
Better on Mainframe
The real tradeoff between cloud and mainframe isn’t cost. It’s agility.
That said, nothing is better than mainframe for highly scalable, secure, transactional workloads. IBM has done a tremendous job in keeping the platform “state of the art” every few years, as it increases mainframe processing and I/O speeds, pervasive encryption, quantum and AI computing and secure storage capacities. That’s in addition to the litany of software tools to modernize data management, AIOPs, DevOps and security.
For over 60 years, it has been the enterprise workhorse where transaction processing (up to 100,000 transactions per second!) is paramount, as is the case in the finance, travel, healthcare, commerce and government worlds.
Immediate Single Source of Truth
With mainframe transactions, there is always a consistent, single source of truth, something that cannot be said of the “eventually consistent” source of truth that the cloud is known for. For highly sensitive or incredibly time-sensitive transactions, this is absolutely vital for many large organizations.
There can be only one credit card or bank account balance, and it better be up to date and accurate. With cloud computing, it is much more difficult to guarantee where nodes become consistent with each other over time.
Horizontal and Vertical Scaling
Unlike on the cloud, mainframe transaction processing can be scaled both up and out (vertically and horizontally). This can be done with no application downtime by adding or unlocking processors and hardware components.
During the COVID pandemic, for example, leading credit card companies found themselves with huge spikes in transactions. But with the help of mainframes, they were able to quickly scale (both vertically and horizontally) their clustering capacities to meet the unprecedented demand.
Because of this, most large enterprises in the finance, travel, government and healthcare industries choose mainframe for its horizontal and vertical scalability.
Hardware-Level Redundancy and High Availability
Furthermore, mainframes “never rest” and have an excellent reputation for never going down. Again, this is why so many highly regulated and large-scale industries entrust their most secure data and transactions to mainframe. It is architected for redundancy and high availability at the hardware level.
Modern mainframes never lose a transaction, even if a disaster occurs during processing. In short, enterprises that leverage mainframes for their critical data can be assured that this data is backed up, secure and completely under their control.
While cloud environments provide disaster recovery, it can be difficult and expensive to guarantee that these environments meet the strict standards of regulated enterprises. This reality makes the mainframe the most resilient platform for mission-critical workloads.
Top-Shelf Security Lowers Relative TCO
The mainframe is also one of the most secure platforms available. Mainframes have always been the gold standard for security and data integrity.
Unlike the cloud that favors open speed, the mainframe favors encrypted security at the hardware level and “zero trust” policies to ensure no one gets access to anything unless they have permission. That’s why mainframe security has stood the test of time, and why it’s the only platform large enterprises will use for their most important applications.
For some IT executives, mainframes can seem expensive. But when the total cost of ownership is examined—volume of data processed, the resilience of the system and the security they provide—the true cost compared to cloud is quite competitive, if not better in large-scale applications. Mainframes are particularly cost effective.
Conclusion: ‘Right Choicing’ for Hybrid Success
There are definitely types of applications that have no business being on the mainframe and should definitely be hosted in the cloud. Web, mobile, service integration and even analytics applications simply make more sense on the cloud. But highly secure transactional enterprise workloads belong on the mainframe.
The real risk of mainframes in today’s rapidly changing digital world is not their cost, but their agility. Most mainframe applications are written in older, less modularized languages (like COBOL or PL/I). It can be hard to understand these legacy applications or find the skills needed to maintain, enhance and operate them.
Because of this, decision-makers must remove these agility inhibitors (lack of skills, application complexity and old or outdated tools) to win on the mainframe. They must allow developers to use modern tools to understand, develop, secure, test and operate applications on the mainframe. They must train their staff on the latest mainframe technology, both hardware and software.
Either way, the most successful hybrid IT environments are all about “right choicing”—reaching for the cloud when web and mobile applications need to be quickly developed, and calling on the mainframe when the transactional backends and data need to be securely processed at maximum speed.