TechNote:
Storage Virtualization
Automated Storage Provisioning
Automated provisioning and storage management, powered by DataCore
Software's SANsymphony™ and SANmelody™ products, offers the ability to consolidate and manage any
heterogeneous collection of storage resources in a hardware and
application-independent fashion.
Pooled Storage Capacity
Storage capacity is viewed as one or more pools that can be
allocated to any storage client using almost any operating system on a
Storage Area Network. Because allocation is from a pool, capacity
utilization efficiencies can be improved by up to 50%. Even high-end
proprietary storage elements (EMC, IBM, Hitachi etc) can be combined in
a single pool, as a tiered storage hierarchy. With lower cost storage
elements making use of arrays based on Serial ATA drives different
classes of storage to be allocated according to the needs of the
application,
Drag-and-Drop Storage Allocations
Drag-and-drop storage capacity allocations to multiple application
servers connected over a mix of Fibre-Channel (FC) or IP networks makes
it simple to address storage growth, changes, and failures without
disturbing the host hardware. The capability of using iSCSI along with
FC also means that different classes of applications can be addressed
with an appropriate host connection, that is independent of the actual
storage hardware connection.
Server deployment is independent of storage purchases. This
"Open Systems" approach protects legacy investments but frees
the user from vendor lock-in. Proprietary storage needs to be
configured only once and then used as a general pool for storage
provisioning. The cost savings in proprietary storage hardware
maintenance calls alone can be a substantial contributor to the ROI for
implementing an automated provisioning capability.
Just-in-Time Storage Provisioning
The use of the DataCore’s Network Managed Volumes (NMV’s) option
allows dynamic, just-in-time provisioning of storage capacity to host
devices. Physical storage capacity is automatically and proportionately
allocated exactly according to real-time usage needs. A “perfect
distribution” of capacity is maintained with no inaccessible islands
of unused capacity. Additional physical capacity can be installed as it
becomes necessary without any interruption of operations.
High-Availability Mirroring, Locally or Remotely
SANsymphony and SANmelody provide the ability to replicate or mirror
storage to another site either synchronously over Fibre-Channel, or
asynchronously over a TCP/IP link, as part of a high-availability
architecture or disaster recovery plan. Unlike proprietary mirroring
schemes that are tied to hardware peers, the DataCore’s mirroring can
be between dissimilar storage hardware, and the modest and sensible
licensing costs are per-site, rather than per-device. See the InfraStor
Tech Note on Asynchronous Mirroring
for more information.
Snapshot Automation
Real-time snapshot automation enables LAN-free backup from a
centralized backup server, eliminating the impact of backup traffic on
host devices, and reducing the costs of backup software licenses.
Repetitive snapshots as a part of a business continuance architecture
can be configured to reduce the impact of failures that occur between
formal backup periods.
Auto-Discovery on the SAN
SANsymphony and SANmelody discover prospective storage clients,
storage devices, and interconnecting switches, automatically giving the
IT staff a complete live picture of their storage network. The software
assigns descriptive names to these assets. The active displays help
isolate cabling and other hardware issues reported to the centralized
alarm console and event manager.
Unsurpassed Performance on Commodity Hardware
SANsymphony and SANmelody use a sophisticated caching algorithm
coupled with a N+1 scalable architecture to provide unsurpassed storage
performance on commodity storage hardware.
- In a recent benchmark of storage performance
audited by the
Evaluator Group,
transaction rates as high as 400,000 IOPS (Input/Output Operations
per second) and data throughputs of as much as 2.1 GB/s were
sustained. This represents 2 to 10 times the performance of any
comparably configured high-end storage subsystem.
- In another independently conducted, SPC-1 performance
benchmark,
DataCore’s architecture demonstrated a standardized Cost per SPC-1
IOPS that is 1/2 the cost of the lowest priced mid-range storage
subsystem, and 1/3rd to 1/5th the cost of high-end proprietary
arrays.