Storage management with Chris Poelker - Part I

1. You can break out SAN management into the following tasks:

• Discovery/mapping
• Policy creation/management
• Device management (LUN/RAID creation, deletion, expansion)
• Subsystem management/tracking (versioning, firmware, maintenance)
• Zoning
• Chargeback/allocation monitoring
• Performance monitoring/tuning
• Quota management
• Event/error management
• Alarm/alert tracking and reporting
• Backup/recovery
• Replication/DR management
• Documentation updates, topology updates
• Labeling, Fibre/cable management
• Capacity planning
• Disaster recovery planning/testing
• Scripting and automation generation

The three most time consuming of these is usually performance tuning, capacity planning and documentation. Most good storage management tools have the capability of providing statistical histograms, which are chartable for both performance and capacity. The hard part is to review these statistics and provide guidance and planning with the application/operations and business staffs.

There needs to be consensus from the business side with budget allocation and the technical application side for performance and capacity needs. Coordinating and advising on these efforts can be a time consuming task. This is where a SAN management team seems to spend a lot of time in meetings. Developing the proper recommendations by reviewing and analyzing all this data can take time, and must be done with input from all affected. This is one of the reasons moving control of data management and access to a separate central SAN group is essential in proper SAN management and coordination. This is a new paradigm for many in the applications groups who normally controlled storage attached to their individual servers.

On the documentation side, all labels and SAN topology diagrams MUST be updated to reflect changes during the lifetime of a SAN, especially SAN with multiple large fabrics. Wiring diagrams and cable labels are essential for proper troubleshooting. This takes time, even when the software uses self-discovery, because software cannot label cables. Mislabeled cables can cause nightmares and even outages during normal maintenance. I usually recommend a CMS (cable management system) for large SAN environments.