On Jan. 19, 2038, an IT problem threatens to have a major impact worldwide. Where did the problem come from, what will happen in 2038 and is there a solution to it?
The 2038 problem actually dates back to the early days of the computer age. Back then, 32-bit processors were the standard. Although these processors were a logical choice then, they threaten to cause problems in the future because they run into the limitations of the Unix timing format. The Unix timestamp is used for keeping track of time. Although this timestamp was devised for the Unix operating system, it is also very dominant on Linux systems. Linux has a huge presence on all kinds of crucial devices. The Unix format ultimately records time as the number of seconds since Jan. 1, 1970. It indicates how many seconds elapsed from midnight on Jan. 1, 1970.
All 32-bit processors run into the limitation of the Unix timestamp. A 32-bit number can represent a maximum of 2^31 – 1 different values, which means that a computer with a 32-bit processor can represent a maximum of 2,147,483,647 seconds. This maximum will eventually be reached in 2038, so there is not enough space to store time.
This limitation means that after Jan. 19, 2038, at 03:14:07 UTC (04:14:07 Dutch time), computers with a 32-bit processor will no longer be able to record time correctly. An overflow occurs when the system tries to create a numeric value that it cannot register. Some computers will crash, but on Stack Overflow there are also stories of negative time logging. Systems then think it is 2^31 seconds before Jan. 1, 1970. That equates to December 13, 1901 at 20:45:52.
How big is the problem?
The 2038 problem is a time bomb for systems. If nothing is done, several companies will be greatly affected. Organizations that rely heavily on accurate timing measurements and continue using a 32-bit processor will run into problems because they use Linux and Unix heavily. This is conceivable, for example, in the aviation industry, which relies on time-sensitive systems for navigation and communication. Should timekeeping fail, there could be potentially dangerous consequences for the safety and operation of the aircraft.
Financial services and healthcare providers are other industries where the 2038 problem is problematic on paper. In financial services, timekeeping is crucial for processing transactions. If time information is incorrect, errors and disruptions can occur. In healthcare, timekeeping is often essential for diagnoses and treatments. Without the correct information, patients are at risk.
For regular users, the 2038 problem is less of an issue. Consumers are not directly very dependent on timekeeping. For them, a system with a 32-bit processor after the 2038 date will mean they cannot use it or see the time incorrectly. Significant problems may only affect them when using company services, as outlined above.
What is the solution?
In the tech world, the 2038 problem is well known. Consequently, many companies have taken measures to get ahead of the problem. The most logical one is a switch to 64-bit processors. Systems with such a processor will eventually have no problem with time notation for a long time. The maximum number of seconds that can be recorded amounts to hundreds of billions of years. This is so far into the future that it is inconceivable that a 64-bit processor will ever run into a limit such as the 2038 problem. In fact, the 64-bit time format exceeds the lifetime of systems and software.
The market is also used to 64-bit; software and operating systems have been adapted to use this architecture. It makes a 64-bit processor by far the most logical choice to address the 2038 problem.
In addition to this logical switch to 64-bit, one solution may be to choose an alternative time representation. This requires complex work by a developer creating elaborate data structures for the year, month, etc. This option is not practical and is preferred in very few situations.
Wake-up call
With the year 2038 approaching, the world seems ready for the challenge. Doomsday scenarios seem to be out of the question as plenty of measures are being taken by companies. Besides the fact that in 1970, it was technically challenging to go beyond 2038, it was not necessarily foreseeable that computers would become an all-important part of our lives for a long time. Now, systems must be adapted to continue meeting society’s needs.