In a database mirroring configuration the primary database must ship the transaction log to the secondary over the network TCP/IP link. In an asyncfullpage configuration the primary will ship the transaction log page to the secondary and normally not wait for a response before proceeding. BUT in both async modes the primary will periodically send a special request to the secondary and wait for a response to ensure that the secondary is not getting too far behind (read more details in my answer to your previous question). I am suspecting that the "hang" that you are seeing is caused by the primary waiting for this ACK.
You did not mention which version you are using. Some changes were made in recent versions of SQLA 17 that changes when this handshake between the primary and secondary (e.g. addition of a 'lagtime' option). These changes may help?
If your network is experience a high rate of packet loss then I would recommend that you spend your effort making improvements to your network! There is no SQLA mirroring configuration that will work well in a high-package-loss network. If you cannot fix your network then perhaps you should investigate using a different method of moving transactions to a secondary server: live backup and mobilink are two such methods.
HTH
Note: Mary's answer to your previous question provides some details on configuring your mirror system.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.