The Storage Economist

In the past I have written about the correlation to costs (savings) and activities to reduce the costs.

For example:

• Thin provisioning can reclaim disk space, thereby saving CAPEX costs in the near-term
• SSDs can reduce the kWatt and BTU per TB (at an I/O level), thus reducing environmental costs
• Archive systems can reduce the backup costs and effort, and many more

There is another dimension that we can introduce between costs and technology, and that is the relationship to service levels. Let’s review some examples:

• Very fast provisioning (mean time to provision) can be enabled with dynamic provisioning
• Automated tiering that is policy-based, can move data (host or application un-aware) to yield the best placement on storage that gives the best price while meeting user requirements
• When data becomes stale, policies and data movers can take over to place the data in a tier that is less expensive foro the company, and when the activity rate increases, the data can be promoted to a faster tier of service
• Backup systems can have variations based on price, RTO, RPO and frequency that meet the customer data protection needs
• Workloads with predictable and measurable performance needs can be kept at the right level, while unknown or unpredictable loads can be relegated to different tiers of service aligned to the required I/O or throughput

An agile data center implies a level of abstraction away from one vendor or one array, and to a higher view of services and capabilities. More people are looking to define service level-based storage capabilities, and are using a storage service catalog to communicate and enable this transformation. Within the SLA, money has to be an option for the user to understand the cost of the service he or she is receiving. Agility will occur when costs (not price) are used in conjunction with service requirements to deliver an IT resource. The type of disk, network, and server behind the “curtain” is not relevant. The customer pays a rate, and is given a level of service commensurate with the price paid.

Just as well the underlying technology to enable the service has to be abstracted from the end user. Whether an application is running on a dedicated server, or a virtual machine; a thick physical storage device or a thin, virtual volume - should not matter as long as the service definitions and parameters are being met.

Using storage economics, service catalogs and storage architectures we can map our innovations and technologies such as virtualization and dynamic provisioning to our customers’ service level objectives and KPIs, leading them to an agile data center with lowered operational costs.