In the fast-paced race for technology innovation, a business can't make it to the finish line without first laying down the groundwork.
There are three components to innovation, according to Dandy Weyn, worldwide compete strategy leader for data and AI at Microsoft, speaking at Talend Connect last week:
Combining data with artificial intelligence and machine learning
Scalability on the cloud
Businesses can struggle to possess a single pillar, and building up all three and establishing the foundation for digital transformation and innovation poses an unwieldy challenge. Each area is fraught with challenges, such as legacy infrastructure, budgetary constraints and talent shortages.
Weyn and other IT executives at the conference in New York City offered lessons, tips and tricks for building out an effective data management system, cloud capabilities and analytics tools. With the right platforms, software and tools, these three pillars offer companies the scalability, agility and power for continued transformation.
Data quantity is nothing without quality and access
The confluence of new technologies, such as the IoT and sensors, and the tools to handle the data they produce, such as the cloud and machine learning, means that companies are handling more data than ever before.
What will determine the winners from the losers is who puts the right data to work the fastest, according to Mike Tuchen, CEO of Talend. Companies need to liberate data from underlying infrastructure so data can be used wherever it needs to be.
The first step is cultivating good data. Measures such as GDPR are forcing data accountability on many organizations, and despite the short-term headaches it is causing, in the long-run companies will be more attuned to what data they collect, store and use.
Pulling data from sources and pooling it in a data warehouse is an important first step, but businesses need to also structure that data and ensure proper access. Access has been a hard spot for most companies because when data isn't filtered or structured it is more difficult for employees to use. It also means more employees have access than is necessary.
Instead of opening the entire data pool up to a marketing or HR employee, a business would be better served having those employees only see the data that is relevant to their area of work. This reduces potential misuse or improper access of data while also streamlining the work of individual employees.
With a common data platform and good data management system, companies can use data more efficiently within the business and build value off of it with analytics, AI and ML.
The tools of the trade
In the past, software was driven by the technology industry. But in the coming years, the other 85% of the global economy will drive this growth, ushering in new platform demands and buttressing the growth of microservices, according to Steve Singh, CEO of Docker.
Software development within industry verticals opens opportunity for niche, specialized applications over broad services and products available across the board. And as individual sectors and companies build out their software skills, they are also building out AI and ML capabilities.
"Data by itself is nothing without artificial intelligence, and artificial intelligence is nothing without data," Weyn said. By 2020, data is expected to grow to 44 zettabytes, and AI investment is shooting up with it, already up 300%.
Even with proper data management, having enough of the right data to train algorithms and derive insight can be challenging. Open data sets are helping beef up image recognition and malware detection capabilities, but for the average company AI skills and resources are still lagging.
More time and data will help companies catch up, but in the meantime political and societal fears need to first be assuaged.
The dialogue surrounding AI and automation has seen a growing emphasis on the integration of automated technology with human capabilities — not as a replacement to humans. But the initial friction is already being felt, and the rest of the pack may have a long way to go before catching up to the pioneers in the space.
Multicloud and hybrid cloud capabilities will rule
While they work for some, on-premise data centers don't have the scale or capabilities to handle the demands and data of the future, and 80% of organizations are adopting a cloud-first strategy, according to Weyn.
The cloud allows for the separation of storage, compute and metaservices, according to Matt Glickman, VP of customer and product strategy at Snowflake Computing. Yet getting there isn't quick and easy: it can take one full week to move a petabyte of data from on-prem to the cloud — and petabytes of data are increasingly becoming a norm in the enterprise.
Few companies are comfortable relying on a single cloud, so the ability to run workloads on multiple clouds is important. While most companies won't necessarily run the same job on multiple clouds, the ability to seamlessly move jobs between platforms with minimal friction is important.
If a company is writing code to one platform, then its IT strategy is that platform, according to Singh. Modern companies need to be more strategic.
In a keynote at Talend Connect, Tuchen and his colleagues demonstrated the capabilities of serverless containers live, showing attendees that they too can build and run jobs across Microsoft Azure, Amazon Web Services and the Google Cloud Platform without having to start over from the beginning each time. In addition to pay per second billing, greater scalability and portability and simple maintenance, serverless and containers will maximize resource utilization.
After all, the cloud gets expensive — and fast. Many attendees and speakers at the event discussed the surprise that can be tied to the arrival of a cloud bill. Even with customer alerts to help reduce spending in areas of low utilization, services on AWS and other cloud platforms can quickly rack up a high bill.