Hadoop Learning notes-which scenarios are not suitable for which scenarios

Hadoop is not a panacea, some scenarios apply, and some do not.

The purpose of Hadoop design consists of the following aspects, namely the so-called application scenario:

1: Extra Large file

Can be hundreds of m, hundreds of t this level of file.

2: Streaming data access

Hadoop works for a single write, multiple-read scenario, where data is copied and analyzed over a long period of time.

3: Commercial Hardware

That is to say the street everywhere can buy the kind of hardware, such hardware failure rate is high, so have a good fault-tolerant mechanism.

Next talk about the scenarios that are not applicable:

1: Low Latency data access

The purpose of Hadoop design is high throughput, so there are no optimizations for low latency data access, and if low latency is required, you can look at hbase.

2: A large number of small files

Because Namenode stores the file's metadata in memory, a large number of small files generate a lot of metadata. In this case, the number of millions other documents is still feasible, and then there are a lot of problems.

3: Multi-user write, Arbitrary modification

Hadoop does not now support multi-person write, arbitrarily modified features. This means that each write is added at the end of the file.

Hadoop Learning notes-which scenarios are not suitable for which scenarios