Hi, Your question would likely be better suited in the Cassandra mailing list.
In a few words: you can, in a way, consider ZooKeeper as a database, as it persists data. But the features used are mostly the synchronisation services to manage the cluster state. So the question to your first question is 'no', so the issue mentioned in the second does not occur in practise. You can have a look at the Google paper on Big Table, it explains its role in a typical BigTable implementation. Nicolas On Tue, Sep 18, 2012 at 12:14 PM, Tuan Luu <[email protected]> wrote: > Dear, > > I'm new to Zookeeper and I'm still not sure how it works with database > system. > > Thus I have some questions and I hope you can help me with them. > > 1. Is every server in Zookeeper system also have a copy of database? So > every time a server starts, if will connect to database and download > everything to its local memory?Or each server in Zookeeper system is also a > database itself? > > 2. What would happen if size of database is bigger then disk size of > Zookeeper server? > > 3. Is it possible for me to connect Zookeeper system to my NoSQL Cassandra > database? I really don't understand how the system will work! When a read > request comes, what is the execution flow between Zookeeper and Cassandra > database? > > Thank you very much. > > *Kind Regards,* > > *Tuan, Luu Dinh* >
