The basic distributed architecture of Elasticsearch (study notes)

1. Elasticsearch's transparent hiding of complex distributed mechanisms.
    Elasticserch is a distributed system. The distributed system hides complex distributed mechanisms in response to large amounts of data.
    Fragmentation mechanism: When we insert data into Elasticsearch, we don't care how the data is fragmented, and in which shard the data is stored.
    cluster discovery: cluster discovery mechanism, when we add a node, the cluster will automatically discover and add the node to the cluster.
    Shard load balancing: es will automatically distribute shards evenly to maintain balanced read and write load requests on each node.
2. Elasticsearch's vertical expansion and horizontal expansion.
    Vertical expansion: increase the configuration of existing nodes to achieve the effect of expansion.
    Horizontal expansion: adding new nodes to achieve the effect of expansion.
3. When adding or reducing es cluster nodes, the data will be rebalanced to maintain load balance.
4. The master node The
    master node does not carry all requests, so it will not be a single point of bottleneck.
    The master node mainly manages the metadata of the es cluster, such as index creation and deletion, maintenance of index element data, node addition and removal, and maintenance of cluster metadata.
    By default, the cluster will automatically select one node as the master node.