No longer tied to a monolithic enterprise price tag, many of the sophisticated storage management capabilities outlined below can now be found in affordable SAN midrange systems from Compellent, iQstor, Xiotech, and other vendors.
Automated data migration allows data to be moved from expensive first- or second-tier storage to cheap SATA-based storage, without manual intervention by the administrator. Automated data migration helps ensure that the highest-performance storage is reserved for the most active or most valuable data.
Automatic volume expansion allows for a thin-provisioned volume to grow automatically when it reaches a threshold. If the volume reaches, say, 80 percent of capacity, then additional capacity -- as long as it's still available on the system -- could be provided automatically.
Boot from SAN allows for servers to boot from the SAN rather than internal disk. This, combined with snapshots and mirrors, allows for one server configuration to be copied and used for a number of physical servers. Since the copies can be made very quickly, a single configuration can be used to boot hundreds of servers in a very short period of time. Combined with automated management tools and server virtualization software, boot from SAN allows administrators to have a pool of server hardware that can very quickly boot any of a variety of operating systems and server software to meet changing requirements.
Content addressed storage (CAS) provides a way for data to be archived and retained for a set period of time without being changed. Data is stored using a hash algorithm that verifies data integrity and proves that data hasn’t been modified since it was stored. CAS is used primarily for meeting requirements for legal discovery and archiving requirements of standards such as Sarbanes-Oxley. CAS is not (yet) available on any of the midtier systems I have tested.
Continuous data protection (CDP) is a journaled synchronization process, usually asynchronous, that allows more control than simple replication. Because each change to each item that is replicated is added to the journal, it is possible to find and restore a specific version of any file that has been changed multiple times, providing a degree of granularity not available with replication alone.
iSCSI support allows the system to provide storage via either the FC (Fibre Channel) interface or an Ethernet interface. Systems that support remote replication can also easily support iSCSI, since both use (and require) an Ethernet interface and TCP/IP capability. The Compellent, iQstor, and Xiotech SANs I tested all support iSCSI. The Xiotech and Compellent systems included the hardware and software necessary, while the iQstor system included the software but not the required hardware module.
Mirroring is another name for local replication. Mirrors can be created, and then broken, so that two initially identical volumes can then be used for different purposes.
Replication comes in three varieties: local, asynchronous remote, and synchronous remote. Local replication keeps two local volumes synchronized with the same set of files. Asynchronous remote replication is typically done over an IP network connection, synchronizing data between a local and a remote volume at specific intervals. Synchronous remote replication typically keeps a local and a remote volume synchronized at all times. Due to the relatively high bandwidth required for synchronous replication, it is typically done via long-distance FC rather than via an IP connection.
Snapshots are point-in-time replicas that are made to use for backups of open files. A snapshot can be made very quickly, and it reflects the state of all data at the time of the snapshot. Because a snapshot is static, a tape backup can then be made from the snapshot without worrying about open files.
Thin provisioning allows for the creation of volumes that can be expanded without deleting the old partition and creating a new one. Thin provisioning can also save considerable storage because it allows for the creation of a relatively small partition that can be grown whenever it gets close to being full.
Tiered storage is used to set up policies that keep critical data on the fastest storage. Typically, first-tier storage uses 15,000 rpm (15K) FC drives, second tier uses 10,000 rpm (10K) FC, and third tier uses 7,200 rpm (7.2K) FC. The fourth tier likely consists of the much less expensive SATA-based storage, which can be up to 10 times cheaper than FC drives. Xiotech offers an even faster first tier with its optional SSD (solid state disk) drives; SSD drives are very, very high performance but also very expensive and low capacity (6GB).
Virtualization means that volumes visible to servers can span multiple partitions or enclosures in the physical storage system. Virtualization provides more flexibility, both during the initial configuration of the system and when expanding storage capacity. For example, an additional shelf of drives can be plugged in, the storage formatted, and the available storage allocated to new or existing volumes.