Скачать или смотреть Kostas Tzoumas, dataArtisans -

01. Kostas Tzoumas, dataArtisans, Visits #theCUBE!. (00:21)
02. Give Us A Little Background On The Company. (00:47)
03. Why Is Stream Processing So Topical Now. (01:29)
04. What Really Made Continuous Processing Take Off. (03:07)
05. Give Us An Example Of How You Would Use Stream Processing. (04:28)
06. When Should You Use Flink Versus Splunk. (06:16)
07. Do We Need To Rethink How We Build Applications. (08:16)
08. Tell Us What We Should Expect From Flink. (09:51)

Track List created with http://www.vinjavideo.com.
--- ---
To Flink or Spark? That is the question for stream data processors | #BigDataSV

by Marlene Den Bleyker | Mar 31, 2016

Apache Spark has been getting a great deal of buzz lately as being the answer to analyzing Big Data, but the scoop at #BigDataSV is that Apache Flink may be a better alternative. The team at data Artisans GmbH are using Flink to improve batch data processing because they like the real-time processing and low data latency of the open-source platform.

Kostas Tzoumas, cofounder and CEO of data Artisans, sat down with George Gilbert (@ggilbert41), host of theCUBE, from the SiliconANGLE Media team, at BigDataSV 2016, where theCUBE is celebrating #BigDataWeek, including news and events from the #StrataHadoop conference. The conversation surrounded the debate about “Flink or Spark?”

Raising capital

As startups go, data Artisans is having a great year. The company started about one-and-a-half years ago when it uncovered the gap for a high-performance stream processor in the marketplace. The company recently announced that it has raised about $6.5 million in a Series A financing round led by Intel Capital, with participation from Tengelmann Ventures and an existing seed investor b-to-v Partners.

Stream processing popularity

When asked why stream processing is so topical now, Tzoumas said, “If you ask me, I wonder why stream processing hasn’t been around for a long time.” He explained that stream processing enables the obvious, which is continuous analytics on data that’s being used continuously. “The most interesting datasets out there are not static; there are events coming up all the time to look at user behavior and click strips, information from sensors, so the very logical thing is to do the analytic on the continuous datasets constantly,” he added.

He believes that the technology is driven by data from the Internet of Things (IoT) and companies looking to make real-time decisions. If you wonder why this is taking off, Tzoumas will tell you, “A lot of companies have been doing this for a while. Banks and telcos and the rest of the world is following.” He said that it is becoming mainstream in the enterprise, and one of the key factors is the maturity of the open-source software for stream processing.

Flink for heavy data lifting

In the debate between Flink versus Spark, Tzoumas admitted he is biased, but he noted that both Spark and Flink cover a lot of types of analytics and each project has strong points. One of the points he made is that Flink is a true stream processor with low data latency on event analytics. So you can have an event and it is flagged in milliseconds. Additionally, Flink takes into account that the streams you are receiving do not arrive in order.

Flink is not for companies looking for a mere add on; if stream processing is at the core of your business infrastructure, Flink is your solution, Tzoumas said.


@theCUBE
#BigDataSV #StrataHadoop