Just started reading the Documentation of Zookeeper. Read that zk has servers ( followers + leader) and clients. Who actually are the clients of zk ? The nodes of distributed system that it co-ordinates ?
Also read that
ZooKeeper applications run on thousands of machines, and it performs best where reads are more common than writes, at ratios of around 10:1.
Does this means that znodes are thousands in numbers ? And what kind of read and write do we want on zk ?
Who actually are the clients of zk ?
A client is any process that connects to the ZooKeeper ensemble using the ZooKeeper client API. Apache ZooKeeper ships with API bindings for Java and C. More information about the Java API is available in the JavaDocs and examples and recipes.
ZooKeeper applications run on thousands of machines, and it performs best where reads are more common than writes, at ratios of around 10:1.
Does this means that znodes are thousands in numbers ?
The "thousands" here refers to the number of machines running ZooKeeper, not the number of znodes stored in the ZooKeeper ensemble. A znode refers to a node stored within the ZooKeeper cluster's hierarchy of data, similar to the concept of an inode in a tradtional file system.
And what kind of read and write do we want on zk ?
Reads refer to operations that get data from znodes or set watches to be informed when changes are applied to znodes. Writes refer to operations that create new znodes, delete existing znodes, or change data attached to znodes.
Reading through the API docs, examples and recipes should shed more light on all of this.