Behaviour of ASM mirroring and disk latency

Hi all,

I would like to understand clearly the behaviour of ASM mirroring when a failure gruop experience a latency. Actual configuration:

• 2-nodes cluster (nodeA and nodeB)
• 2 sites (siteA and siteB)
• In oracle definition is extended cluster (campus distance)
• Each node has 4 paths per LUN
• Each diskgroup is created in 'NORMAL REDUDANCY' with two failure groups
• Each failure group is made by many LUN
• Each node has a preferred_read_path set to its own site.
• Linux multipath is used

I want to understand now, from a database point of view, which is the behaviour if an OLTP database (single instance) running on nodeA write on diskgroup DATA01 and there's a latency in connection to siteB? AFAIK ASM needs to perform a syncronous write so log writer waits ASM successfully write on both site. As siteB is reachable with a latency ASM does not consider the failuregroup lost. Hence, the logwrite experience a wait proportional to the highest latency site.