Article ID: 17387
Product: Exchange
Date: 27 December 2006

Title: Good Mobile Messaging server for Exchange - failover methods and cache storage options

Question: What options does Good provide for server failover methods and cache storage?

Answer: There are a few configuration methods for disaster recovery when a Good Messaging server goes down, all of which are contingent on Exchange being functional. If Exchange is okay and the Good server’s cache folder is intact then there shouldn’t be any user impact. If the cache is lost or corrupt then Good can still be recovered but all users will likely need to recover wirelessly by pressing "r" to reconnect. when automatically prompted. While high-availability Good server cache replication methods are being developed, most of our customers find the following recovery methods sufficient.

Automatic (Warm) Failover: A primary and standby Good Messaging server in a Microsoft cluster configuration with the Good server cache residing on either a shared external drive or SAN. If one Good server fails, then the clustering service automatically transfers the primary role to continue processing the cache and message queue to the standby node. Admin Guide Chpt 9

Manual (Cold) Failover: A primary and standby Good Messaging server in a non-cluster configuration with the Good server cache residing on either a shared external SCSI drive or SAN. If one Good server fails, the admin can then delete the cache lock file and start the services on the standby server.  For more details please contact Good Technical Support and request the PDF document from KB 17197.

Manual Install Failover Method: A primary Good Messaging server and a designated host that any Good Messaging server configuration may be moved to if needed. The primary server's cache may be stored on any supported system and if available after a failure would be copied to the new host. This is Good's most popular option even among larger accounts due to it's flexibility and low cost.

- If the previous Good server is available then the admin can copy over the cache files and have a seamless recovery by following the standard "move to new hardware procedure". Please review: Admin Guide Chpt 7 Sect 1-2 or KB 13025.

- If the previous Good server is unavailable or the admin cannot copy over the cache files to the new host, then the admin can still recover the configuration; however, all handhelds will likely receive "Error 49: Disconnected from server" from which they can wirelessly recovery by pressing "r" to reconnect when automatically prompted. Please review: KB 17532.

Cache Storage: Good currently supports three storage types: local e.g., a drive in the server, attached external e.g., a shared SCSI array between two servers, and SAN (storage area network) drives, i.e., a special network designed for storage. Good does not currently test or support any real-time file replication technologies such as Doubletake, XOSoft, SRDF (Symmetrix Remote Data Facility), or Volume Shadow Copy. NAS (Network Attached Storage) is not supported. These solutions can actually cause damage to the cache files if they try to lock them while in use. Good is evaluating high-availability cache replication solutions including site to site SAN replication.