Communication with Cloud DFS is a big bottleneck.

[taeguk]

A connector supported by Cloud DFS is blocking until communication is finished.
So, it can be a big bottleneck in a slave which uses the connector.

Related issue : DrawML/cloud-dfs#7

This issue can occur lazy reactivity and expiration of slave.

[taeguk]

can't process other message when process one message.
But enabling processing several messages parrallely can occur big synchronization issue.
And this is very heavy job for now.