4 ms·
I'm not entirely sure the point the author is trying to make. As an honest question as someone who's not even played with it, why would hadoop ever get past 1%
by mattmanser 14y ago
I'm not entirely sure the point the author is trying to make. As an honest question as someone who's not even played with it, why would hadoop ever get past 1% or probably even 0.01%? Doesn't it have a very specific use case?
Is he only talking about penetration in the BigData market? If not, why would it ever supplant RDBMS or NoSQL datastores?
- sounds 14y agoI'm obviously the wrong person to talk about Hadoop - having so little experience with it - but this article strikes me as link bait. To answer your question, the author seems to think if you have Hadoop, you'll never need NoSQL. Here's his quote: "I think we're going to see Hadoop produce the next Oracle-sized database company. We're also likely to see such a company emerge from the NoSQL ranks, but Hadoop is a near certain bet right now." This quote also seems wrong to me: "In fact, if anything, we are still only on the first few steps of the Big Data marathon, a race that Hadoop seems set to win despite its many shortcomings." I believe the Big Data industry is well on its way. There are multiple big players offering an array of cloud services and software.
- lmm 14y agoHadoop is the first NoSQL datastore to reach a level of maturity and big-name support that many enterprises insist on. In these companies the only thing for it to supplant is traditional relational databases, and the advantages there should be obvious.
- mattmanser 14y agoI don't see any obvious advantages of NoSQL in most enterprise apps. Although it's such a fuzzy term these days.
- monstrado 14y agoI wouldn't think of Hadoop being a "specific use case", because it's really just a platform that you design around. Of course, everything needs a use case, but Hadoop's place in the industry is not nearly as linear as people think it is. The problem at the moment is people don't understand or know how to work with Hadoop, and that's what companies like Cloudera and Hortonworks are trying to solve, they are trying to show businesses, 1) How easy it is to deploy and 2) How to actually apply it to their current datasets. As someone who works at a Hadoop company, I can tell you that customers are still trying to wrap their heads around the technology, but the good news is, there is major progress. I notice that, rather than Hadoop not giving enough to the user, the user is usually not giving enough to Hadoop. Hadoop challenges its users to think differently, which for some businesses can be quite difficult. Regarding your question about NoSQL, since Hadoop is a platform/framework, there are already efforts to design databases that take advantage of the stack (namely, HDFS <-- Hadoop Distributed Filesystem). The most popular of these contributions is HBase which is a NoSQL (KV) Column Based DB, based on BigTable and much like Cassandra. It's horizontally scalable and is often referred to as "highly consistent". Just to make a point though, there is so much that runs on the Hadoop platform. Top level projects like * Pig (Scripting language developed by Yahoo, think SQL meets Python) * Hive (Map RDBMS style schemas to data after it's been stored, let's you query data with SQL) * HBase (NoSQL column based database store based on Google's BigTable, think Cassandra but very consistent) * HDFS (Highly scalable and consistent distributed FS that was designed with MapReduce in mind) ^^ Some people use this JUST for storage since it's so easily scalable. * There are others like Mahout which is machine learning, and in my opinion, black magic :P Hope this helps :D