Friday, November 30, 2018

Podcast: Hadoop, JRuby, Grails, and Python Creators Talk Tech Trends

The first thing you may notice when listening to this program is that the podcast has undergone another name change. What was once the Oracle Developer Community Podcast is now the Oracle Groundbreakers Podcast. A little change is good now and then, and this is exactly that, a little change.

And it is by no coincidence that change is the core theme of this program, how various trends and technologies have shaped the IT landscape over the past year, and how other trends will shape the future.



As befitting their varied specialties and interests, each of the panelists offers a unique perspective on the swirl of technologies that have changed, and will continue to change, the software development landscape.

During one segment I asked the panelists to talk about the trends or technologies that have had the greatest impact on them as individuals and on the work they do. 

Graeme Rocher cites GraalVM as having had "a massive impact" on how he thinks about the way in which modern frameworks and Java tools are built.  "If you want to support Graal VM’s capability to compile down to native images, which, again, further allows you to optimize startup time and reduce memory consumption, you really have to plan ahead in terms of how you can make that happen. It's not something you can just add on after the fact. So supporting Graal’s native image has changed my workflow," he admits. "Now I'm considering, should implement this feature? Will it work on Graal? And it's it's having a massive impact on planning in terms of the next 18 months." 

By his own admission, Guido van Rossum lives in a different environment. "My workflow has very little to do with what Java developers typically encounter." In his world, social media has had the greatest impact. "It has spilled over from being social to affecting my work, affecting the Python community. The state of how people are trying to influence developments through social media has really changed recently." 

Charles Nutter spends his time in the tech trenches. "I work pretty low-level. On JRuby I mostly do optimization compiler work, sitting and staring at assembly dumps all day. In the past year his work was most affected by the new LLVM-based JIT compiler in Azul’s Zing, and by OpenJ9 and the availability of the J9 source code. "And then, of course, the Graal JIT, which is separate from the Graal VM project, is actually available as an experimental JIT in Java 11," Charles explains. "You can just flip it on and get the benefit of a whole bunch of new optimizations. It actually helps JRuby quite a bit. So it kind of seemed like we'd gotten to a point where the JVM JITs had gotten as good as they were going to get, and then everything changes again. So it's an exciting time for me on the JRuby project."

In contrast to Charles Nutter, Doug Cutting tends to be more very high level. "I'm talking to people more about the data systems they're building" he says. According to Doug, the technologies that get people excited don't necessarily reflect what they are actually doing. People may be talking about machine learning or artificial intelligence, but that talk doesn't necessarily indicate action. "They’re not even moving to the cloud and in a big way." But Doug sees people are starting to use  and get value from "next-generation data platforms that have been around now for a decade." 

"We really do see people ingesting large amounts of what they call unstructured data, and exploring it using a suite of open-source tools," Doug explains. "That was the hot technology five years ago, and now it's becoming mainstream and beginning to have a large impact in a lot of conservative industries," such as banking, telco, automotive, and healthcare. "That's exciting." 

The excitement and energy extends into Siddartha Agarwal's world as well. "There are two or three things that we've had to focus on quite a bit," he admits.  "We've been focused a lot on delivering a managed Kubernetes service and then delivering a functions platform delivered as a service that is not locked into a particular cloud," he explains.  "We've launched that as an open source project called Fn Project.

APIs have also occupied Siddartha's focus. "APIs are absolutely critical. Everyone's using APIs for everything. So how can you enable them to run the security of the API anywhere?" What about rate limiting policies? What about security authentication? "You need a gateway," says Siddartha. "That gateway must be able to in our public cloud or in on-prem data centers, because you might not have APIs going to the cloud or consumption in the cloud, or running in third-party data centers. So it’s the notion of hybrid, in that it’s more about multi-cloud and across on-prem in public cloud."

No comments:

Post a Comment