Investments in cloud computing will continue aggressively in 2017, but many organizations will opt for multi-cloud environments in their data centers. That is the prediction of Avinash Lakshman, chief executive officer at Hedvig and creator of Apache Cassandra.
Avinash spoke with Information Management about what the New Year will hold for data and IT professionals. He sees five key trends for the cloud and storage markets.
Multi-cloud commitments will become the new normal
“With many companies making investments in public and private cloud services, 2017 will see more businesses committing to multiple cloud providers at the same time,” Lakshman predicts. “For example: there will be fewer and fewer Amazon Web Services-only businesses; rather dual-source public cloud services will be used instead to avoid vendor lock-in. The challenge will come in making data services easy and productive across multiple clouds. Without this function, enterprise deployments will be as inefficient as they were when they were using tape.”
In-memory and temporary storage become more important
Lakshman notes that augmented and virtual reality, artificial intelligence and machine learning have become increasingly popular. “While analyzing these new sources of data is becoming critical to long-term business goals, storing the data long term is both impractical and unnecessary when the results of analysis are more important than the data itself,” Lakshman notes. “Although 2017 will see plenty of data growth that will require permanent storage, most net new data generated next year will be ephemeral; it will quickly outlive its usefulness and be discarded. So despite exponential data growth, there won’t be as much storage growth as we might otherwise have expected.”
More Content Delivery Networks go Do-it-yourself
“Content delivery networks are now a major source of pain and cost because of inaccessibility and low performance,” Lakshman says. “The alternatives have become one of the expensive, complicated provider solutions, or build your own. Expect more companies to build and operate their own CDNs using public cloud and software-defined infrastructure. These DIY CDNs won’t be turnkey, full-service solutions, but they will provide leaner, less expensive solutions to stale, traditional architecture.”
Machine Learning will become de rigeur in the enterprise
“What’s unique to today’s machine learning technology is that much of it originated and continues to be open source,” Lakshman explains. “This means that many different products and services are going to build machine learning into their platforms as a matter of course. As a result, more enterprises will be adopting machine learning in 2017 without even knowing they’re doing it because vendors are actively using ML to make their products smarter. Even existing products will soon use some variety of machine learning that will be delivered via an update or as an extra perk.”
More companies will monetize metadata as new revenue streams
“The nature of many distributed systems, such as those used in large companies such as Google or Facebook, is by design to collect and store lots and lots of metadata,” Lakshman says. “This ‘data about the data’ will become increasingly relevant as companies analyze more of it for insights. Organizations such as Netflix have already built their success around analyzing customer data for commonalities and many companies will follow. Making sense of metadata, particularly metadata that has been stored for a long time, can also lead to new customer insights as well as become the focus of new products sold by analytics vendors.”