Matthias Eckermann (below, right), director of product management for SUSE Linux Enterprise at the the Nuremberg-based company, said in response to queries from iTWire that software-defined infrastructure would bring about a change in existing business processes, and allow new business processes to be implemented.

But he said this did not necessarily mean that hardware businesses were staring down the barrel at extinction.

"There will always be requirements which go beyond 'standard' (think of specific performance or latency requirements!), and for those, specialised hardware and specialised processes may remain in place."

Asked whether he agreed with the proposition that software-defined processes were slowly posing a threat to hardware device makers' profits, Eckermann said: "Let me answer this way: yes and no."

Elaborating, he added: "On the one hand, we do see a 'commoditisation' in the area of standard hardware and standard appliances. (But) on the other hand, however, we see three other trends, which may positively influence device makers' profits:

• "IoT increases the number of devices, device interconnects, aggregators, and required space;
• "In the areas of HPC and machine learning we see a specialisation in the compute engines; and
• "Those systems which are required to be really 'mission- critical' or 'business-critical' may remain unaffected by the SDI trend for the time being."

Given that failures were more common in software than hardware, Eckermann was asked whether the incidence of failure would rise as software came to play a bigger role. "I am optimistic here, as there are two directions to solve this challenge: Software supporting 'mission-critical' deployments and cloud native applications," he replied.

He pointed to his own employer as an example. "[In the case of] software supporting 'mission-critical' deployments, this is the 'classic' stack: SUSE Linux Enterprise Server; SUSE Linux Enterprise High Availability including Geo Clustering; and SUSE Linux Enterprise Live Patching.

As far as cloud native applications go, Eckermann said these consisted of many (partially) independent (micro) services, with one or more instances of each service running.

"This type of application stack takes 'failure' of one or multiple instances into account; thus a hardware defect or a software error can be 'hidden' from the end customer."

Looking ahead a decade or two, Eckermann said there were multiple trends which were overlapping and which would influence each other.

"We have an increasing demand for IoT devices, which are driven by Linux, and specifically a need for stability, security and long-term maintainability.These capabilities — stability, security, long-term maintainability — are interestingly enough the same values which drive the 'classical' data centre and 'mission-critical' deployments today.

"In other words, my first expectation is that the IoT world will learn from today's 'mission-critical' world and adopt their processes and tools for quicker success," he said.

"Cloud native application stacks will (try to) become 'independent' of the hardware and underlying infrastructure pieces. This will drive adoption of software-defined technologies, as I described. Thus my second expectation is that software-defined infrastructure and cloud native application stacks will grow."

Eckermann said interestingly, the trend for even stronger mission-critical needs could be seen even today. "This may sound like contradicting what I said earlier, but I do not see it as contradictory, but complementary," he added.

"The reason is (to make it short) that more information coming into a company increases the amount of business data to compute and store. It also increases the dependency of the business on that data, as the data becomes the business value. As soon as data is a business value by itself, it will neither be corrupted nor lost, and it becomes 'business-critical'. Business-critical data, though, needs 'mission-critical' IT systems.

"So, we could say: not despite the growth in cloud native and software-defined solutions, but because of it, the growth in the true 'mission-critical' world will continue as well."

Eckermann had a one word answer when asked what industry at large needed to adapt to this change: "Trust."

"Security and confidentiality are the big challenges, at least from my (European/German) perspective. Don't get me wrong: this is not about the software 'per se'. We have well established quality and security standards for the software," he said.

"The challenge is more that without any 'piece of hardware' defining the boundaries of a security realm, we need to find new ways to define, implement, and monitor access rights, locality of data."

He said that while there were so-called "attestation" frameworks, interestingly enough they all needed some kind of hardware ("crypto token" or "security zone" for example) to be fully functional.

"That's another example of 'specialised hardware' which we most probably will not get rid of," he said, adding, "I am aware that this is a bit 'cryptic', but that's the best way to express my view on this topic."

In summary, Eckermann said there was high-quality software available to establish software-defined infrastructure which would enable new business processes.

The responsibility now lay with people in the industry to create and establish trust frameworks around this, he said.