PUREcluster Parallel NAS Storage
The PUREcluster Parallel NAS Storage Cluster™
accelerates time to results for enterprise computing clusters. By combining a parallel
file system with object-based storage, the PUREcluster Parallel NAS Storage
Cluster delivers scalable bandwidth and random I/O to accelerate application
throughput and streamlines operations within single scalable namespace
to accelerate productivity.
PUREcluster Parallel NAS Data Sheet
PUREcluster Active Scale Data Sheet
The PUREcluster™ Parallel NAS object-based
architecture from Verari Systems™ offers a breakthrough capability:
it dynamically distributes workload across a multitude of small, intelligent
blades. PUREcluster Parallel NAS StorageBlade™ and DirectorBlade™
are perfectly tuned to work with and deliver outstanding performance
from the PUREcluster Parallel NAS ActiveScale File System. As a result,
the ActiveScale Storage Cluster™ delivers near linear scaling of performance
with increases in capacity. Built from industry-standard hardware
components, the sleek, hot-swappable blades slide easily into a 4U
shelf, storing up to 5 terabytes (TBs) per shelf.
STORAGEBLADES
StorageBlades contain all application and user data and handle the
vast majority of file system activity. As powerful as a stand-alone
computer, each StorageBlade balances two Serial ATA disk drives with
a processor, cache memory, and network I/O to optimize your file system
activity. The PUREcluster Parallel NAS Storage Cluster can easily
be scaled-in capacity and performance- simply by adding more StorageBlades.
This payas- you-grow model means you won’t have to purchase a small
system you’ll eventually outgrow or a system too large for your current
requirements. The blades have the intelligence to manage all the detailed
work associated with transferring data to and from drives—from prefetching
to smart data layout-without requiring help from a file server as
in traditional network storage approaches.
Intelligent StorageBlades work in concert with the PUREcluster Parallel
NAS DirectFLOW™ software enabling a direct data path to clients
(Linux clusters, application servers, or desktop PCs). As a result,
very large files can be accessed much faster through simultaneous
reads and writes to multiple StorageBlades. Each StorageBlade delivers
up to 70 Mbytes-per-second peak throughput, nearly as much as an entire
shelf of disk drives in competitive systems-at the same or lower price-per-gigabyte
as individual Fibre Channel and SCSI drives.
DIRECTORBLADES
PUREcluster Parallel NAS DirectorBlades orchestrate key activities
between clients and StorageBlades. Leveraging the PUREcluster ActiveScale
File System, DirectorBlades virtualize data across StorageBlades enabling
the system to be viewed as a single, easily managed namespace. Each
DirectorBlade contains up to 4 gigabytes (GBs) of high-speed cache
RAM and a Gigabit Ethernet connection to optimize system performance.
DirectorBlades can be clustered together to support highly scalable
storage networks. DirectorBlades simultaneously support the PUREcluster
Parallel NAS DirectFLOW, NFS and CIFS data protocols, enabling you
to configure the PUREcluster Parallel NAS Storage Cluster to handle
huge bandwidth requirements out-of-band, while still supporting record-breaking
performance for standard NFS and CIFS clients. DirectorBlades automatically
manage load-balancing-moving data from heavily accessed blades to
lightly loaded ones-freeing administrators from timeintensive reorganization
activities.
In an easy, granular fashion, you can continuously add DirectorBlades
to the cluster to boost availability, and balance performance and
capacity growth for your specific application needs. StorageBlades
and DirectorBlades are the hardware foundation of the premier storage
system for scalable Linux clusters. Built on an object-based architecture
and optimized to meet the unique needs of Linux cluster customers,
the ActiveScale Storage Cluster is designed to fundamentally enhance
customer expectations for storage performance and management. It offers
breakthrough performance-in throughput and random I/O-exceptional
scalability, and greatly simplified management.