Address Space Manipulation Techniques

This section explores the address space manipulation techniques used in SAN virtualization and volume management. Among the topics covered are

• The storage flat address space

• Aggregation

• Subdivision

• Substitution

• Mirroring

• Private data areas and metadata

The Storage Flat Address Space

Storage (block) address spaces are referred to as "flat." This means they have a simple one-dimensional structure with a beginning address, an ending address, and all other addresses in between, in order. There are no rows or columns or tuple address locations to figure out. Figure 12-8 represents a very simple 20-block storage address space.

Figure 12-8. A Simple, Flat, 20-Block Address Space

Aggregation with Striping and Concatenation

Storage virtualization can aggregate blocks from different sources to create larger address spaces. Figure 12-9 shows a 25-block virtual address space created by combining a 20-block address space with five blocks from another 20-block address space.

Figure 12-9. Concatenation of Storage Address Spaces

The process used to aggregate blocks in Figure 12-9 is an example of concatenation. Concatenation does not intertwine the downstream address spaces; it appends one address space to another.

RAID striping, described in detail in Chapter 9, is a special case of block aggregation where all array members contribute the same number of blocks to the upstream address space. RAID intertwines the downstream storage address spaces by associating strips from the various member drives to form a segment of the upstream address space.

Subdivision

Subdivision allows an address space to be segmented into smaller units. Figure 12-10 shows an eight-block address space and a 12-block address space created from a single physical 20-block address space.

Figure 12-10. Subdivision of Storage Address Spaces

Subdivision allows storage to be purchased in volume to reduce costs and then be parceled out as needed to applications. If the need for storage capacity is known in advance, it may be simpler and less expensive to purchase a large amount of storage all at once. Storage virtualization enables both approaches: buying in bulk or paying as you go.

Substitution

Storage virtualization can substitute one storage resource for another, transparently to the host systems using it. This substitution can be done with different devices/subsystems or address spaces. One common example of virtual storage substitution is using disk storage as a substitute for tape storage in backup, a topic that is explored more in Chapter 13, "Network Backup: The Foundation of Storage Management."

Mirroring

Mirroring, described in Chapter 8, "An Introduction to Data Redundancy and Mirroring," is a special case of storage substitution that takes two identically sized storage address spaces and presents a single, identically sized upstream storage address space. When a failure occurs that keeps one of the mirrored pairs from working, the other continues to fulfill the storage function.

Private Data Areas and Metadata

Virtualization products often have private data areas or metadata that are not part of the exported address space that is used by file systems and databases. Private areas or metadata are for the exclusive use of the volume manager or SAN virtualization systems and are used to store historical information about product operations and configurations.