Nomad_Soul - Fotolia
Exchange Server depends heavily on storage. Understanding what's required for Exchange 2013 storage and choosing the best options improves mailbox server performance.
After you've looked at some ways to tackle sizing problems in your Exchange 2013 setup, you can move on to the best RAID configuration -- typically RAID 5 or RAID 10 -- for how your Exchange setup runs.
Although Exchange 2013 doesn't require RAID, it is highly recommended. IT architects and administrators can use RAID 5 or RAID 10 configurations to boost storage performance, guard against disk failures and provide enough storage capacity for Exchange 2013 tasks.
RAID 10 is most common for Exchange 2013 deployments. It provides redundancy by mirroring the contents of one disk to another. It combines mirroring with striping, which interleaves storage across multiple disks. Striping boosts performance by allowing multiple disks to work simultaneously. Each disk holds at least 256 KB blocks of the operating system, swap file, mailbox database, log files or other content.
RAID 5 and 6
RAID 5 stripes data across a group of up to seven disks for better performance. But instead of mirroring for resilience, it spreads parity data across each disk in the group. If one disk fails, the failed disk's data can be rebuilt from the parity data across the remaining disks. RAID 6 incorporates a dual parity scheme across its group disks, allowing a RAID 6 group to recover from a simultaneous two-disk failure. Rebuilds from parity data can take time and render Exchange unavailable until the rebuild is complete. RAID 5 and 6 may be deployed with additional protection such as mirroring (RAID 50 or 60), allowing the mirrored set to take over until the rebuild is finished.
The array controllers RAID 5 and 6 use should invoke advanced features. For example, surface scanning and data scrubbing are two error detection and correction technologies for important storage groups where proactive diagnostics help forestall disk errors. Both techniques basically read and verify all blocks on each disk, isolating and rebuilding any bad blocks from mirrored or parity data, depending on the basic RAID level.
Exchange 2013 supports just a bunch of disks (JBOD) group storage, but it's not recommended because JBOD groups are logically a single disk with no inherent resilience. It's possible to use JBOD groups if you maintain two or more redundant database and log copies. Exchange 2013 also supports multiple databases per volume on JBOD groups, which can improve storage use; but multiple copies are required to guard against disk faults.
Consider the cache settings for your Exchange 2013 storage subsystem. For example, configure the cache space for either RAID 10 or RAID 5 to support 75% write and 25% read cache. This ratio might change with different Exchange server roles. For example, servers with the client access server role specialize in authentication and redirection and might use 25% write and 75% read cache settings. If the server uses a direct-attached storage system, the write cache may be set to 100%.
The storage array controller cache should also be protected by a battery backup or system-wide uninterruptable power supply. Otherwise, disable disk write caching to prevent storage data corruption if you lose power before cache data can write to disk.
Part one: Avoid storage sizing issues
Storage considerations when virtualizing Exchange 2013
How storage virtualization gives RAID a boost