phoenix JDBC connection Strings - Zookeeper connection(client) VS phoenix query server(thin-client)

[#1]

Hi ,

As we knew , we can connect to phoenix/hbase in 2 ways .

1 . phoenix query server using thin-client , this way we can pass only 1 phoenix query server ip/host to the connect string .

2. zookeeper jdbc - we can pass all zookeeper ip's as part connection URL .

1.What is the difference between them ?

2. Which one can give more performance ?

3. How can we can get load balance if we go with phoenix-query server JDBC (thin-client) connection ?

4. which one is the best method to get read performance ?

[#2]

"What is the difference between them ?" -- The former uses the Phoenix Query Server (PQS) as a proxy to HBase, reducing the client-side complexity, while the latter talks directly to HBase and ZooKeeper. We call the former the "thin" driver and the latter the "thick" driver.

"Which one can give more performance ?" -- The thick driver because there is no extra "hop" in executing a query.

"How can we can get load balance if we go with phoenix-query server JDBC (thin-client) connection ?" -- Read https://community.hortonworks.com/articles/9377/deploying-the-phoenix-query-server-in-production-e.h...

"which one is the best method to get read performance ?" -- The thick driver.