IT departments are always looking to cut costs wherever possible, and that’s as true in finance and banking as in any other industry. CIOs at financial institutions and banks are not unfamiliar with the challenge of making small budgets fit their needs and are often the first to take the hit when budgets are trimmed, making their job even more of a balancing act.

That doesn’t seem always seem fair, considering the ever-increasing workloads that IT systems are constantly asked to handle, and the constantly growing reliance that technology now plays in daily banking operations. In fact, to cut costs across the board even further, many financial institutions are turning to automation and AI, and new technologies like blockchain, which can all streamline banks’ systems and drastically reduce their cost of doing business.

What isn’t always apparent, however, is that these same technologies are taxing IT systems even more - and that poor system performance can end up putting a huge dent into any savings.

Look at blockchain, for example. Blockchain has caught on as a tool for streamlining the way banks do business and process transactions; its database architecture is efficient, and an attractive option for improving many of the processes that banks already have in place – such as the already-complicated clearing and settlement process. Many of the world’s largest banks and exchanges – led by the Australian Securities Exchange - are switching to blockchain for their clearing and settlement systems to cash in on its efficiency – a move that could save them $10 billion a year.

Blockchain’s efficiency and security is a good match for processes such as payments and settling syndicated loans. Credit Suisse and other 18 other banks have already formed a consortium to replace their current clearing systems for syndicated loans with blockchain, in hopes of speeding up the settling process (which can take an average of 19 days).

These improvements will be huge for banks, but the massively-increased burden blockchain places on already-taxed systems ends up being counterproductive; blockchain is extremely data-intensive and requires a massive amount of reading and writing from the CPU for every operation. Add such data-intensive processes to the workload an IT system is already responsible for, and overloads, bottlenecks and system crashes are inevitable.

The same is happening as banks’ increase their use of AI and automation. AI is often expensive to implement up front but can end up saving institutions dramatic amounts by reducing the number of employees needed, and by optimizing how efficiently banks operate.

Improvements in sheer processing power and access to cloud storage make it possible for banks to automate many tasks that previously couldn’t be, including customer support, fraud detection and claims, insurance management, predictive analysis and even virtual financial assistants. Some experts even estimate that within the next 10 years, the number of people employed in the banking industry could be reduced by as much as 30% - 50%.

The potential financial upside from eliminating those jobs is huge, but the more banks turn to AI and automation, the more they’re going to be asking their IT systems to handle.  

The amount of data analysis the financial industry requires is already often too much for even the most well-funded and supplied systems. Most large IT operations and large databases are run on Microsoft SQL server, which – while reliable, powerful, and capable of creating very large databases – does have its own limitations, and is very often the bottleneck that drags down system performance.

In addition, since approximately 80% of the world’s computing systems run on the Microsoft Windows operating system, there is another hidden, looming bottleneck that will become more obvious as huge amounts of data are processed. The way Windows reads and writes data, storage is separated from computing, and thus the system writes data “logically” instead of physically to storage – breaking I/O data (any information that moves between storage and the CPU) down into smaller, more random, fractured chunks. The result is a huge amount of unnecessary I/O traffic (data fragments and noise) that can reduce performance by 30% to 40% or more.

That’s why many systems are so overworked, despite often having the bandwidth needed for handling most of their processes; they just don’t use it efficiently. Adding AI and blockchain processing might save in some areas but will just create worse bottlenecks and crashes. Unless CIOs and banks find ways to keep their performance up while keeping costs low, they won’t be able to use those savings as needed.

The traditional remedy for poor performance is upgraded hardware, but those aren’t an effective solution - especially not if trying to cut down expenditures; the money spent ends up canceling out much of the savings that AI or blockchain creates in the first place. And, while you may eventually have to shell out for new hardware several years down the road, hardware won’t solve a problem caused by software in the first place.

If simple inefficiency in I/O is costing systems and SQL databases 30% or 40% of their performance, then making I/O reads and writes more efficient - through software and not hardware upgrades – would be the most logical and effective way to reclaim that performance. That way, AI and blockchain can be effective cost-savers for banks and institutions, and not a burden.

Jim D’Arezzo has a distinguished career in technology that started at IBM and has included senior executive positions at Compaq, Autodesk and as President and COO of Radiant Logic. He is currently CEO of Condusiv Technologies, the world leader in software-only storage performance solutions for virtual and physical server environments.