The pattern of cloud adoption moves something like the ketchup bottle effect: You tip the bottle and nothing comes out, so you shake the bottle and suddenly you have ketchup all over your plate.
That’s a great visual from Frank Munz, software architect and cloud evangelist at Munz & More, in Germany. Munz and a few other leaders in the Oracle community were interviewed on a podcast by Bob Rhubart, Architect Community Manager at Oracle, about the most important trends they saw in 2017. The responses covered a wide range of topics, from cloud to blockchain, from serverless to machine learning and deep learning.
During the 44-minute session, “What's Hot? Tech Trends That Made a Real Difference in 2017,” the panel took some fascinating detours into the future of self-programming computers and the best uses of container technologies like Kubernetes. For those, you’ll need to listen to the podcast.
The panel included Frank Munz; Lonneke Dikmans, chief product officer of eProseed, Netherlands; Lucas Jellema, CTO, AMIS Services, Netherlands; Pratik Patel, CTO, Triplingo, US; and Chris Richardson, founder and CEO, Eventuate, US. The program was recorded in San Francisco at Oracle OpenWorld and JavaOne.
The Cloud's Tipping Point
The ketchup quip reflects the cloud passing a tipping point of adoption in 2017. “For the first time in 2017, I worked on projects where large, multinational companies give up their own data center and move 100% to the cloud,” Munz said. These workload shifts are far from a rarity. As Dikmans said, the cloud drove the biggest change and challenge: “[The cloud] changes how we interact with customers, and with software. It’s convenient at times, and difficult at others.”
- Related: Try Oracle Cloud for free
Security offered another way of looking at this tipping point. “Until recently, organizations had the impression that in the cloud, things were less secure and less well managed, in general, than they could do themselves,” said Jellema. Now, “people have come to realize that they’re not particularly good at specific IT tasks, because it’s not their core business.” They see that cloud vendors, whose core business is managing that type of IT, can often do those tasks better.
In 2017, the idea of shifting workloads en masse to the cloud and decommissioning data centers became mainstream and far less controversial.
Blockchain: Practical, Enterprise Use Cases
Blockchain is perhaps best associated with Bitcoin (about which there were many jokes during the podcast), but its uses go far beyond digital currencies. Blockchain can be used whenever there’s a need for a distributed ledger that must be trusted and hard to hack—
anything from tracking vegetables through the farm-to-shelf supply chain to integrating applications and data within a huge multinational corporation.
The panel was cautiously excited about the emergence of blockchain, and of tools for using blockchain in cloud applications, but they were wary of its computational cost. “Public blockchains are very expensive because you have to prove your work,” Dikmans said. Adding new data to a public blockchain can require processing and sign-off among hundreds or thousands of nodes, to ensure that there is consensus that the data is trustworthy.
However, not all transactions require such extreme level of trust. Take data that’s entirely created and consumed within a single enterprise, or a small group of partners. Such private and consortium blockchain applications can utilize the same core technology, but require far fewer nodes to gain a trust consensus. Look for these kinds of new blockchain use cases this year.
Serverless Computing
“I like to joke it’s really Stored-Procedure-as-a-Service,” said Chris Richardson. “This idea that you just hand over your code to some infrastructure, and it just runs it in an event-driven way, and if you want you can route HTTP requests to it. It’s actually really cool.” That’s how Richardson gets his head around serverless computing, where code for a microservice can be handed over to the cloud, and the developer doesn’t have to provision or create containers or virtual machines to run it. The cloud platform takes care of the plumbing automatically.
Serverless computing began gaining some traction a few years ago, but Oracle’s announcements at Oracle OpenWorld 2017 have the potential to be game-changers, the panel said. Not only is Oracle’s implementation of serverless handled through the open standard of the Fn Project, but Oracle offered more advanced plumbing that can orchestrate multiple serverless processes to create complex workflows, and tie them to http requests.
“You don’t even have to worry about the VM and the operating system, and how big a VM you need to provision, and all of that,” Richardson added. “You say, ‘Here’s my code, and now run it.’”
Pay-As-You-Go Machine Learning
Yes, machine learning and other AI techniques have been around since the 1950s. Yes, people have been talking about it forever. And yes, finally, it’s different—because of the computing capacity the cloud can deliver and because of new graphics processing units (GPUs), which have been proven excellent for the sorts of matrix math that AI requires.
These GPUs, available as a cloud service, “let you do accelerated ML and DL, all in RAM, and all within the GPU. It’s taken it to the next level, where you can run large data sets very quickly,” said Pratik Patel.
What’s more, Patel said, “You don’t need to buy any of the hardware. You can just go rent space in the cloud and run large data sets for however long you need to, and then you can move on to something else.” Patel noted that the latest GPUs and graphics boards are increasingly expensive, with supply and pricing driven largely by applications such as digital coin mining. So why buy when you can rent?
Meanwhile, the latest generation of GPUs and chipsets are being optimized for ML/DL applications, with machine learning algorithms like TensorFlow being embedded in silicon. Expect the AI revolution to accelerate in 2018.
Listen to the full podcast to hear more about the hot tech trends of 2017 from the developer perspective, and more about ketchup as well.
Alan Zeichick is principal analyst at Camden Associates, a tech consultancy in Phoenix, Arizona, specializing in software development, enterprise networking, and cybersecurity. Follow him @zeichick.
