With the amount of noise surrounding things like Amazon Go store, self-driving cars, and even the use of home assistants such as Amazon’s Alexa, many people these days have at least a general understanding of artificial intelligence (AI).

So, it’s no surprise that AI has quickly made its way into the core of our everyday lives. However, these overt and obvious uses aren’t the type of cases that today’s IT professionals immediately think of. Instead, they are thinking about the type of AI applications that can have an impact on practically everything.

Some of the biggest technology challenges in today’s cloud-first environment include things such as IT outages, ransomware and security breaches, as well as being able to seamlessly and securely migrate data to and from an environment. But what can AI applications do for them?

IDC’s recent The State of IT Resilience report states: “The emergence of non-traditional data types requires innovative backup and recovery methodologies.

“Application data, machine learning data, and data gathered from sensors – ranging in format from structured to unstructured – will all be relevant to an organisation’s IT resilience strategy, creating an ongoing data management and visibility challenge.”

Essentially, by combining backup and disaster recovery (DR) technologies with AI, IT teams can begin to address some of these challenges and make a big difference to the IT industry. The most likely application for this use is machine learning (ML), as although AI allows computers to mimic human intelligence, ML is the modest, unassuming sibling that can mean machines improve tasks through experience.

The issues when it comes to migrating

When it comes to a business’ data use, AI is on the complex side of the scale. IT managers first need to make sure that their IT infrastructure is flexible, robust and secure enough to facilitate it. This kind of resilient IT environment then ensures access to critical applications, with no disruption to either customers or operations – something that becomes even more relevant when we think about the enormous amount of data needed for ML to work.

A prime example of this is Amazon Go, a shop that Amazon is currently testing that uses a combination of camera images, as well as sensor and mobile application data. This means shoppers can take what they need and leave the store – all without having to interact with anyone.

So, if Amazon needed to migrate some of this crucial, ML-reliant data that makes Amazon Go work, it’s critical that they have a resilient IT infrastructure. This can mean the difference between an Amazon Go customer that is unaware a seamless migration even took place, and a customer that is compromised because the migration happened to cause major malfunctions and vulnerabilities.

The possibility of hands-free DR

Being able to continually run necessary failover tests, tests that mean duplicating and moving data over and over again to ensure a DR strategy works in the case of a real emergency is a common challenge many IT organisations face. By using ML-based prediction tools, businesses are able to see projected storage rates, as well as being automatically provided with relevant options.

An example would be ‘what if’ scenarios such as moving data to various places, the costs involved to move the data to different cloud platforms, and how much storage could actually be saved by removing cold data. Additionally, organisations can use ML that’s advanced enough to act on pre-determined preferences – being able to automatically balance storage resources based on algorithms that factor in available cost, previous manual decisions, space and beyond.

• Changes to the European Banking Authority outsourcing guidelines: What you need to know
• The rise of Akari: An interview with Director Margaret Totten
• New Edinburgh consultancy looks to shake-up AI ethics

In an ideal world, any business’ DR system should be able to automatically, seamlessly and efficiently do its job better than what could be done manually should a disaster occur. However, we aren’t quite there yet. Automatic disaster recovery sounds appealing, but the ML still needs to advance to the point that it can ensure, for example, a network is only down for five minutes, and a resource-intensive recovery is certainly not necessary or worth it. Despite all of this, there is still the possibility of a hands-free DR system. A business and its customers
experiencing less downtime is the end-goal, and can be reached by using the right algorithms and variables, as well as enough data to assist machines to determine situations such as waiting five minutes before performing a failover.

With enough algorithm-based trial and error, the machines will eventually perform better, quicker and smarter than humans. It would even enable a DR process that could be so efficient, even an unexpected hurricane hitting a 200,000ft data centre would mean nothing to the business’ data centre servers.

Obscuring the lines between security and recovery

With ML continuing to advance, the line between security and recovery is vastly disappearing. Deep machine learning can and will help security and recovery systems work together to make smarter IT decisions. An exciting and promising way in which AI and IT resilience can come together is when security systems that are equipped with advanced ML can detect a ransomware encryption – as it takes place.

This then means it can automatically connect with a recovery system to both stop the attack as it’s happening, and failover to unencrypted data. This example will likely be realised in the not-too-distant future, and no business will be able to compete or survive without deploying this kind of technology, especially with the rate in which cyber threats are growing in frequency and sophistication.

In the world of backup and recovery, there are likely to be many other potential AI-based scenarios – many of which likely won’t even be discovered for years to come. It’s not uncommon in our evolving, data-driven world to see rapid and unpredictable change. However, one thing that we can say for sure is that the backup and recovery technology will be unrecognisable, even in just a few years’ time.

Technological innovation is changing rapidly, and the implications for how organisations can better preserve and protect data will enable a future where the concept of downtime is unheard of.