I came into the technology market working for a division of IBM in the 1980s. Over the next few years, the company faced intense challenges partially attributed to the belief that the platform that had supported IBM for decades, the mainframe, was dead.
Sun Microsystems was by far the most vocal advocate of this idea that the mainframe — and by proxy, IBM — was dead. They were also the biggest beneficiary of that trend.
Here we are around 30 years later, and Sun has been dead for about a decade. What is left of the company is a badly wounded hardware unit in Oracle. By contrast, the mainframe is still around, and Hitachi just licensed the technology. So not only hasn’t the mainframe experienced its demise, it is moving back to being a multi-vender solution. (I should add that for much of the interim time this platform has remained IBM’s most profitable.)
To put this in perspective, this is kind of like being told your 90-year-old mother is dying. But when you arrive at the hospital, you find out that you have a new sister and your mom has just entered Wimbledon as a contender.
The Pendulum Swings
Trends in the technology industry often work like pendulums, particularly when it comes to where computing is done. Computing started out centralized on mainframes with dumb terminals. Then it went to PCs at the edge connected to networks. Then it went to client/server computers, and then back to a blend of desktops and servers. Next it moved at least partially to smartphones. Currently, it is moving towards the cloud, which is again a centralized service.
You didn’t need a mainframe, which is optimized for massive I/O, if the work was being done on PCs; they could just hit centralized storage resources. But as loads moved back to the center, the need arose for a platform that could support a massive increase in I/O and provide access and search services for ever bigger databases.
Z Systems, the current name for IBM’s mainframe, does a really good job of solving a problem that the industry itself created. You see, for the better part of a decade the industry pushed the concept of big data because it wanted to sell storage.
And it worked — it worked way too well.
Firms and services bought storage like there was no tomorrow and filled it with data, and not just structured indexable data either. They started storing audio files, video files and sensor date that wasn’t just unstructured, it was very hard to index and, as a result, very hard to make sense of. And here, more wasn’t better.
Much as with a needle in a haystack, with unstructured data the more you have, the harder it is to find what you need. Because all this data found itself in huge repositories, everyone and his brother needed access to it. This was a near perfect setup for the mainframe, which has as its main advantage the ability to provide massive access to huge data repositories but is relatively limited in processing power.
By the way, this situation is about to get much worse. Smart hospitals, autonomous driving cars, connected airplanes and ever-smarter cities are expected to explode these massive data repositories when they come online. This is likely partially why mainframes suddenly became a ton more interesting.
In the 1980s, the mainframe wouldn't have been able to address big data needs, but today the z Systems are designed to function like another server, albeit a big one. The processing can be outsourced to another platform, like IBM Watson, which can analyze the data while the mainframe facilitates access.
The Hitachi Opportunity
The one big negative for the mainframe — at the beginning and again recently — was that there was only one vendor: IBM. At first, given that mainframes were sold more like a service, this wasn’t problem because there was no real alternative. But now, people really don’t like to bet on any one vendor for a certain technology. So even though the opportunity for mainframes has been going vertical, the demand was massively offset because of the concerns surrounding single sourcing. This is a particularly big problem for government accounts, which often have hard rules in place preventing relying on a single vendor.
With Hitachi again entering the segment, that problem starts to go away. This is one of those instances where giving, as in granting a license, will likely be better for IBM than receiving. You know, similar behavior, licensing out PC technology, led directly to the PC market and currently forms the basis for smartphone market success.
In the end, the market has come around again. What was old is not only new, but highly relevant and profitable.
Photo courtesy of Shutterstock.