These days, because David J. Dewitt wrote an article on Database column: mapreduce: a major step backwards, many foreign websites have very popular discussions about this post! Both parties have a lot of Daniel from the industry to participate in the discussion. At present, the opposition basically accounts for the majority, and some netizens regard David's document as a joke;
Some domestic websites have also reproduced some of these discussions, but most of them are part of them, so it looks tired and some context is incorrect! Just because of my determination to strengthen the English Reading Training, I also had some understanding of mapreduce principles in the early stage due to project reasons, and I have also seen some implementations of hadoop; so I went straight to the original post and its discussion post: database people hating on mapreduce.
After reading this, I think David doesn't mean he wants to deliberately degrade mapreduce, and he has affirmed it in many parts of his article, he also has a deep understanding of its working principles and application scenarios. For example, mapreduce may be a good idea for writing certain types of general-purpose computations, but to the database community, it is...
However, he made a comparison with rdatabase to solve the problem. The two are a technology unique to different fields and scenarios, comparing them is not a fair thing! As a result, many Google-related fans are starting to get angry ......
From David's point of view, I personally feel that he is still making mistakes in his mindset. He is an expert in the DB field, but that's why, he is always used to his inertial thinking when looking at many things. After decades of constant thinking, he has derived some advanced features in RDBMS, such as schema, index, transaction, views ...... Mr does not take the ordinary path, and it avoids index and other problems when processing large-scale data. David knows this, however, he still needs to maintain the good features of RDB and maintain his achievements over the years;
In addition, Google and IBM will provide a computing environment with 1000 CPUs for many universities, while Berkeley wants to take the Mr programming paradigm as a course, david thinks this will have a negative impact on the student union, because Mr uses a similar brute force to solve the problem! I envy those foreign students who can quickly access some of the most advanced algorithms and computing technologies! In comparison, Chinese students are not so lucky. If you want Mr to enter the curriculum of Chinese college students, I think it may not be a play in five years!
To sum up, I think David is a cool man, but he is too stubborn and stubborn!