| The logical breakdown of storage intelligence and its geographic placement within the overall storage infrastructure mandates categorization. As a starting point, we define six basic storage functions, or services, required to provide enterprise-class storage solutions. Those services include manageability, capacity, recoverability, performance, security, and availability. Note that virtualization is not specifically called out as a service itself. Due to the numerous uses of the term , we employ the classic virtualization definition in our framework. That definition ”to abstract heterogeneous storage homogeneously ”can be part of one or more associated services. That abstraction may be across multiple disk drives , volumes , or storage arrays. However, the associated services, not the abstraction layer, determine the customer value.. Chapter 3, "The Software Spectrum," outlined overall storage management features across three focal points: infrastructure, transactions, and disaster recovery. That framework fits with our storage services, as shown in Figure 4-1. Figure 4-1. Storage services within the storage software landscape.  4.1.1 Storage Service Definitions The basic services in Figure 4-1 define the storage portion of the service. The overall group is therefore focused on lower layer activities that extend up to volumes, but not necessarily including file systems. The services set supports any combination of platforms for block-based , file-based , or object-based storage. Since the services are independent of the storage platform, we can evaluate service offerings across almost any platform or network type. -
Manageability ” As expected, manageability ranks high on the list of required services. With installations becoming ever more complex, management of the storage infrastructure consumes larger portions of the overall operating budget. Users strive for manageability through simplified administration, configuration, monitoring, and repair. Today storage system management spans servers, HBAs, SAN switches, subsystems, and tape libraries. Each can have its own interface and method of management communications, causing unnecessary complexity. -
Capacity ” The capacity service falls into two categories ”LUNs and volumes. Ultimately, the storage service deliverable is a volume made up of LUNs, which in turn comprise one or more disk drives. With simplified SAN deployments, volumes typically reside within a single array or enclosure. Today, with volume aggregation capabilities available in the network, "virtual volumes" may include capacity gathered from a range of drives or arrays. -
Recoverability ” Even with highly available configurations, the potential for data corruption exists. Whether recovering from a dropped packet or a corrupted database, the ability to rapidly return to a steady and operational state determines overall deployment availability. Upon encountering data corruption or system failure, point-in-time copies facilitate one way to return to order. With lower-level recovery issues, such as a dropped packet, protocol exchanges in the networking layer recover accordingly . -
Performance ” Performance services encompass methods to boost storage throughput or transaction benchmarks with available resources. One example is RAID 0 functionality to create volumes. By striping data across multiple disks, the effective I/O throughput increases . Other performance-enhancing measures might be to use 2-Gb/s high-speed Fibre Channel networks to boost throughput between high-end servers and storage. -
Security ” The deployment of a storage network requires access and control mechanisms across the infrastructure. In direct-attached systems, the physical security of the data center may have been sufficient. With wider storage distribution across the enterprise, storage administrators need the ability to actively monitor and control access to valuable storage resources and prevent these resources from being used to further compromise the system. -
Availability ” Protecting environments from sudden configuration changes, high-availability services facilitate rapid changeover to alternate active computing engines, including storage. One common form of availability is RAID 1 functionality, effectively mirroring data across two devices. In the networking world, multicasting provides similar capabilities, allowing one packet to be sent to two or more destinations. By categorizing storage services, we can evaluate sample implementations and placement within the overall storage infrastructure. Implementations vary across hardware and software, but more specifically traverse previously cordoned areas now open with the reach of storage networks. Storage services can be provided through host-based applications running on servers, within network implementations running on appliances or switches, or within target subsystems such as RAID devices. The areas of storage intelligence and potential hardware platforms are shown in Figure 4-2. Figure 4-2. Potential storage services locations.   |
