Learning Machines Go Forward With Backup

This article was written by Alistair Mackenzie, CEO and founder of two UK companies, Silverstring Ltd and Predatar. He as formerly senior storage specialist at IBM after being in sales at Bell Microproducts and Hammer Distribution. Silverstring was founded in 2002 initially as a VAR company but evolved to become an MSP. Predatar was a product formed within Silverstring in 2007 and was launched as a separate independent company in March 2018 as a SaaS business. Predatar works with IT channel MSPs or VAR companies wanting to transform or evolve to stay ahead in a shifting digital world.

Learning machines go forward with backup
The Case for AI in backup and Recovery

The likelihood of any job being automated can be understood by asking the 4Ds; is the job dull, dangerous, dirty or dear? Not sure about dirty or dangerous but people tell me backup is dull whenever I discuss it, and it’s often very expensive to operate.

Our ability to keep pace is also a problem. Keeping data protected is a growing challenge as the volume and variety of data increases. And if data is the new oil, it’s also becoming more of an imperative.

Backup and recovery best practices move slowly from company to company, often only when skilled engineers move jobs. In ecological terms this is the equivalent of DNA mutation through procreation; a very slow evolution.

New data points from IoT and the proliferation of cloud computing models all point to data protection becoming a bigger challenge. With increasing regulation and ever more sophisticated malware attacks, we need to accelerate this evolution to keep up in the data protection arms race. Hackers are using AI techniques to steal your, or your customers’ data, it’s time you considered it too.

Applications of AI in Data Protection
Machine learning is different from regular programming. Rather than developers writing explicit code to perform some computation or task, a machine learning application can use the data to create models.
One use case, increasingly used by the most successful service providers is improving customer satisfaction. An AI powered customer service chatbot can support and scale business teams in their relations with customers.

Predictive analytics can help answer the question of what will happen in the future based on what is happening now. This technology can be used to identify patterns of backup failures and predict when the next one will occur.

How to get the best from AI

The Right Architecture
In 1957 Frank Rosenblatt built the Perceptron, the first trainable neural network algorithm. To help encourage a neural network to learn, you need a large network. In 2007, when Apple launched the first iPhone, Steve Jobs insisted developers use only web apps via the Safari browser. This was a failure because it was a closed ecosystem and inward looking. Only when Jobs allowed an SDK to be published did third party developers stream in, creating an unprecedented wave of innovation.

AI is about asking questions. As no one person has the answers, in AI no one person has all the questions. Apple’s success is built on the principle of a large ecosystem, or network.

For the successful adoption of AI in data protection, we believe that the learning platform must be separated from the underlying applications. Developers must have access to the metadata, so they can ask the questions to turn dumb backup data into smart insights. Large enterprises can use multiple backup tools so having your AI platform out-of-band will create more training data for machine learning.

An Open Mindset
A requirement for the successful adoption of AI in data protection is a large ecosystem. Our second belief is that the service provider and vendor channel community will be at the heart of future AI innovation in data protection. Managed and cloud service providers, as they scale, will turn to AI to solve the toughest problems in backup and recovery. The most successul AI platforms will need to embrace te channel ecosystem.

70,000 years ago, at the time of the first cognitive revolution, Homo Sapiens learnt to change and adapt the stories they were telling. This allowed them to form larger groups and networks. These groups learnt to work together to defeat enemies which were individually stronger than them.

In the middle ages in London, England, merchants and craftsmen formed Livery companies to protect themselves and their customers. Highly skilled artisans, such as the Leather Sellers and the Pewterers, even though competitors, would join together to gain network effects. Today, despite a global communications revolution, clusters still exist. Hollywood is where movie makers congregate, Silicon Valley for tech start-ups, and in Oxfordshire, UK, exists a hub for Formula One teams.

So if we are to apply the same logic to today’s world wide digital age, just as economies prosper when trade is unrestricted, AI in data protection will flourish with companies that provide open architectures which in turn will foster innovation across a wide ecosystem.