Hitachi Unified Storage VM, Руководство по обзору продукта

Страница: 23 / 70

System overview
1–11
Hitachi Unified Storage VM Product Overview Guide
tiers of storage, including devices optimized for archival. Users do not
necessarily need to be aware of the actual data location, and applications
need not be rewritten either
High flexibility and wide applicability
The Hitachi File system flexible architecture enables wide applicability to
changing workloads, data sets, and access patterns. It provides high
scalability, fast metadata processing performance, and excellent data
movement speed under a wide range of host loads, usage patterns, and
data types. Fine-grained parallelism, off-loading of specific file system
operations to FPGAs, and data pipelining all contribute to the Hitachi File
System handling of both throughput and metadata processing.
Separation of server and storage functions
The Hitachi File System delivers performance with relatively small storage
systems. The file system also allows for performance to increase granularly
as more disks are added. Typically this benefit is immediately apparent,
even before "restriping" the data across both old and new spindles, as
writes spread automatically and immediately. As a result, customers may
start small and scale performance by adding storage when needed.
Additional file servers are not necessarily required for additional
performance. As performance requirements grow, customers may also take
advantage of clustering technology within the Hitachi File System to add
more servers while maintaining a single namespace. This provides easy
management of large pools of data, but the Hitachi File System still offers
true separation of function between storage and servers. Each may be
scaled independently to meet your needs; there is no requirement to
purchase one to get the other as with many competing NAS product
High namespace scalability
Scaling beyond a single NAS server is essential for high performance
storage solutions. Many parallel file system implementations rely on
clustering multiple servers together for greater aggregate performance.
Individual servers are much more powerful than traditional CPU-based
architectures, meaning fewer servers are needed in a given cluster to
achieve the same level of performance. The difference with the Hitachi File
System is the scale. The Hitachi File System also makes it possible to create
a single, unified namespace across the entire cluster of file servers. This
makes it appear as a single file system to all network hosts and is known as
cluster namespace or CNS. CNS satisfies the most common scalability
requirements, allowing network hosts to access data on any server in the
cluster, regardless of physical location. The Hitachi File System moves data
seamlessly between multiple cluster nodes with minimal impact to
performance.
Advanced multitier storage mechanisms
The Hitachi File System provides policy-driven data migration mechanisms
for transparent data migration between many storage tiers. Data has an
assigned value (by age, data type, owner, etc.) so transparently relocating