MinIO is the world's leading object storage pioneer, with millions of users around the world. On standard hardware, read/write speeds are up to 183 GB/sec and 171 GB/sec.
Object storage can serve as the main storage layer to handle various complex workloads such as Spark, Presto, TensorFlow, H2O.ai and become a replacement for Hadoop HDFS.
MinIO is used as the primary storage for cloud-native applications that require higher throughput and lower latency than traditional object storage. These are the performance indicators that MinIO can achieve.
Simple and scalable
Compatible with AWS S3 to achieve seamless connection
Install
For other installation methods, please refer to the official documentation.
docker install minio
docker run -p 9000:9000 --name minio1 \
-e "MINIO_ACCESS_KEY=AKIAIOSFODNN7EXAMPLE"\
-e "MINIO_SECRET_KEY=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY"\
-v /mnt/data:/data \
-v /mnt/config:/root/.minio \
minio/minio server /data
Object storage principles
MinIO's encoding method encodes an object into several data blocks and check blocks. We call it Erasure Code for short. This is the type of encoding. This type of encoding also requires algorithms to implement. Minio uses Reed- Solomon's algorithm
Even if you lose half (N/2) of your hard drives, you can still recover your data.
Minio uses Reed-Solomon code by default to split the data into N/2 data blocks and N/2 parity blocks.
With 16 disks, an object will be divided into 8 data blocks and 8 parity blocks. You can lose any 8 disks (regardless of whether they store data blocks or parity blocks), and you can still get data from the remaining disks. Recover data from
explain
B is a matrix of nx (m + n)
D is a real data block that we can prove
Figure 1
When storing, store it as n+m, that is, the amount of redundancy is m
Figure II
At this time, the data D1, D4, and C2 are lost.
Figure 3
According to matrix operations, it can be seen that the original data matrix D can be obtained by the invertible matrix of Survivors matrix *B`
above all
The maximum tolerance for data loss is m
The amount of data redundancy is m
In general, m = n, so it is 2 times redundant.
Bucket notification guide
If changes occur in the bucket, such as uploading objects or deleting objects, you can use the bucket event notification mechanism to monitor and publish them in the following ways