Confusion on Mysql Management node? Where to run queries?

Follow

Drew Hamlett

• July 09, 2012 18:15

Please forgive me if this has been asked before, but I was under the impression that the I would connect to the management node to run queries.  I'm guessing this is not the case though.  Am I suppose to proxy the sql nodes via Ngnix or a HA solution myself?  The Mysql management node is just for managing the other nodes?  Correct?  

I'm just wondering how I would do pooling with a webapp.  Say I connect to one of the nodes directly, does that mean the pool would just be on that  server?  Thanks for any help on this subject.

0

• 

  Johan

  • July 09, 2012 19:12

  Hi,

  This is a great question, many have it so thanks for bringing it up here.

  Management servers (ndb_mgmd) are NOT involved in query handling and no queries/routing or whatever related to queries are relayed through the management serverrs, but it manages:

  • Cluster Configuration
  • Writes a log file (ndb cluster log)
  • .. and a couple of more things (network arbitration etc)

  Query processing is handled by the MySQL Servers connected to the Data Nodes. The MySQL Servers converts SQL to a protocol the data nodes understand (the api is called the NDBAPI).

  You typically have to MySQL Servers (for HA).

  You can LB over these two MySQL servers using e.g HaProxy or a Hardware LB.  Some SQL drivers like Connector/J + php mysqlnd have Load Balancing built in that can be used.

  
  

  Hope this helps.

  
  

  BR

  johan

  
  

  
  

  
  

  
  

  
  

  
  

  0

  Comment actions Permalink
• 

  Severalnines Support

  • July 09, 2012 21:36

  By the way, you can easily add HA Proxy to your MySQL Cluster or Galera deployment using these scripts:

  https://github.com/severalnines/haproxy

  0

  Comment actions Permalink

Please sign in to leave a comment.