Comments on: 5 reasons why the future of Hadoop is real-time (relatively speaking)

By: Brian Stone

Brian Stone — Wed, 13 Mar 2013 08:54:55 +0000

Fantastic article Derrick, full of great content. Two additions from Causata’s perspective. Regarding Real-Time, we call it Web-Time. Causata makes decisions in 50 milliseconds or less to place offers, ads, and related content on web pages, mobile apps and with social campaigns. Our machine learning applications for web and advertising personalization run entirely on an HBase event data store. In addition, we’ve partnered closely with Cloudera to bring our v4 release to market. Mike Olson has repeatedly told Causata’s CEO Paul Wahl that the availability of high-quality, valuable applications on the Cloudera platform is a critical factor in Cloudera’s long-term success. Causata is the very first vendor to put together an end-to-end application focused on an important business problem with Cloudera.

By: viplav

viplav — Sat, 09 Mar 2013 01:36:56 +0000

Great series and great post on Hadoop trends. It is unfortunate that the term Real-time is used so loosely by the database/bigdata crowd while the term was already used to mean guaranteed response within strict time constraints. I had to chuckle when I heard for Impala “real time” is waiting less: “It’s when you sit and wait for it to finish, as opposed to going for a cup of coffee or even letting it run overnight…. That’s ‘real time.’”. Raymie’s terms synchronous and asynchronous sound more accurate – but the marketing people might not like them that much.

By: Geoffrey Hendrey

Geoffrey Hendrey — Fri, 08 Mar 2013 06:50:21 +0000

The main takeaway from strata this year was “shorter time from HMMM to AHA!” Hadoop needs to be faster. Real-time was the mantra. My notes from strate keynotes:: http://vertascale.com/faster-time-from-hmmmm-to-aha/

By: Sam Taha

Sam Taha — Fri, 08 Mar 2013 02:13:44 +0000

Good points made here about where Hadoop is going as it expansion beyond batch oriented. Related article about Hadoop subject here: http://grandlogic.blogspot.com/2013/03/sql-and-mpp-next-phase-in-big-data.html

By: anon

anon — Thu, 07 Mar 2013 18:37:58 +0000

Derrick, you mention “why HBase has attracted such a strong following despite its relative technical and commercial immaturity compared with comparable NoSQL database Cassandra. “. Why would you say this? HBase runs at Facebook at a pretty impressive scale. What makes it immature?

By: timmoreton

timmoreton — Thu, 07 Mar 2013 16:57:55 +0000

Derrick, this is a great post. I would agree that “real time means different things to different people”. Here at Acunu we focus on real-time operational intelligence on high velocity streams, and our typical customer architecture sees data collected via Apache Flume being sent to both Acunu Analytics and a Hadoop cluster — Acunu for real-time decisions to be made on metrics that are known to be important, and Hadoop for mining new trends and correlations on data at rest. Timely answers for operational intelligence relates to the decisions that need to get made on the back of the metrics; whereas for investigative analytics it’s more about interactivity, and keeping up with “speed of thought”.

I wrote up some more thoughts on the topic here: http://www.acunu.com/2/post/2013/01/the-different-meanings-of-real-time.html

-Tim