02-05-2014, 01:04 PM
Network Support for Network-Attached Storage
Network Support for Network.PDF (Size: 127.08 KB / Downloads: 39)
Introduction
Storage systems represent a vital market with storage
densities growing at 60%/year, resulting in 35%-50%/year
decreases in the cost per byte. In recent years, the amount of
storage sold almost doubled each year and is expected to
sustain annual growth of at least 60%. Secondary storage
has a healthy place in future computer systems.
While many storage products are directly attached to
personal computers, most disk array products (65% and ris-
ing) are deployed in local area network file servers. This
centralization of storage resources enables effective sharing,
better administrative control and less redundancy. However,
it also increases dependence on network and file server per-
formance. With the emergence of high-performance cluster
systems based on commodity personal computers and scal-
able network switching, rapidly increasing demands on stor-
age performance are anticipated. Specifically, storage
performance must cost-effectively scale with customer
investments in client processors, network links and storage
capacity.
NASD Implementation
To experiment with the performance and scalability of
NASD, we designed and implemented a prototype NASD
storage interface, ported two popular distributed file sys-
tems (AFS and NFS) to use this interface, and implemented
a striped version of NFS on top of this interface
[Gibson97b]. The NASD interface offers logical partitions
containing a flat name space of variable length objects with
size, time, security, clustering, cloning, and uninterpreted
attributes. Access control is enforced by cryptographic
capabilities authenticating the arguments of each request to
a file manager/drive secret through the use of a digest.
Network Support for Network-Attached Storage
The success of the NASD architecture depends criti-
cally on its networking environment. Clearly, support for
high-bandwidth, large data transfers is essential. Unfortu-
nately, traditional client-server communication paths do not
support efficient network transport. For example, measure-
ments of our NASD prototype drive (running DCE/RPC
over UDP/IP) show that non-cached read or write requests
can easily be serviced by modest hardware.
CONCLUSION
High-performance, low-latency networking is essential
to achieving the potential of scalable network-attached stor-
age. User-level networking solutions, such as VIA, have the
potential to do this, but must be mindful of the amount of
on-drive resources required — connection state and buffer-
ing can consume considerable resources. However, Remote
DMA can help minimize drive resources while providing a
great deal of flexibility in drive scheduling and buffer man-
agement. Further, VIA’s application-level flow control
enables aggregation of flow control across arbitrary storage
components, something low-level network flow control is
not designed to support.