|
The Potential of IP Storage
By David M. Piscitello, Core Competence, Inc.
Originally published by Interop This Week, reposted with permission.
Two key indicators tell me that networked storage is a
mature technology. The first is that most everyone in the industry can explain
the difference between a storage area network (SAN) and networked attached
storage (NAS). Immediate recognition of an industry acronym is always a signal
that a technology is on everyone’s radar, and it’s been some time since anyone
concluded I was asking about the San Diego and Nassau airports, which share
these same acronyms. The second is that storage has embraced IP and Ethernet.
And no other combination of communications technologies is a clearer signal of
a march towards ubiquity and commoditization than “IPoE”.
More than storage…
Our information compulsive society has an insatiable
appetite for storage, and it is the one appetite technology is currently able
to appease, affordably. Bandwidth is woefully lacking everywhere but the local
area and core networks. Operating system and office application upgrades assure
that we never have enough CPU and memory. But we can afford storage aplenty.
The problem however, is not simply how much storage we have, but our growing
inability to efficiently manage the vast and disperse repositories we’re
storing. Networked storage-or more precisely, storage management applications-
address this problem in several ways. They provide the ability to efficiently
archive and restore information, eliminating the waste incurred by storing and repeatedly
storing, let alone archiving, dozens if not hundreds of copies of the same
information. This, in turn, creates efficiencies in cataloging information,
assuring information accuracy, facilitating search and retrieval or timely
restoration. IP-networked storage also offers certain efficiencies and improved
scalability, redundancy, and diversity in server (farm) architectures by
introducing the ability for servers to communicate with storage over
potentially long distances.
Perhaps most overlooked and singularly important however, is
the improvements in accountability and auditing networked storage may yield.
Current events and new legislature remind us that all organizations will be
held more accountable for information they have been charged to safeguard.
Improving accuracy alone makes networked storage useful in industries where
information is regulated, e.g., health care, but consider the luxury of having
confidence that only the approved and “signed-off” copies of documents are
preserved – in a law suit, the fact that you know exactly what will
surface during the discovery process is an invaluable asset. We are fast
approaching a time when organizations large and small will recognize that it’s
better to know exactly what information they have stored and the legal
ramifications of holding such information than to be rudely awakened in a court
of law following a discovery process.
The fact that even small organizations can accrue the
benefits of networked storage makes the “IP” in storage most significant.
Familiarity breeds Content
Nothing
disrupts any operations staff, be it small or large, than the introduction of
radically new technology and infrastructure. What IP brings to storage, and the
reason it’s destined to succeed, is familiarity. By deploying iSCSI, for
example, SMB’s can create networks of storage devices using familiar Ethernet
technology. Hosts and servers using TCP/IP can reliably access these. The TCP
connection is essentially acts like storage bus in a block storage (SCSI) I/O
operation. This simple explanation alone will set many an administrators mind at
ease – “I know SCSI, I know IP, I know Ethernet – kewl…”
At the risk of being criticized for painting the rosiest of
pictures, it seems that the path to commoditized hardware will come quickly for
iSCSI, and that first generation performance will soon be improved by TCP
offload engines and more highly integrated NIC and storage host bus adapters. I
can’t help but believe that storage management software prices will follow, or
that SMB suitable storage management software will come to play.
“IP” storage comes in different flavors, and can deliver
benefits to large enterprises as well. Large enterprises can create TCP/IP
tunnels to connect geographically distributed Fibre Channel anywhere
they can find adequate bandwidth to satisfy storage application requirements,
using the soon-to-be Fibre Channel over IP (FCIP) Internet standards-track
protocol. There is also a
standards-track protocol (iFCP) for connecting Fibre Channel storage devices or
SANs using IP instead of Fibre Channel switches. Organizations in metro areas
where Gigabit Ethernet service is offered can look to IP storage as a means of
satisfying business continuity requirements. While technology constraints
currently hamper such deployment, Gigabit services may eventually extend this
reach nationally and internationally.
New Service Opportunities
It’s too early to predict whether IP storage will rejuvenate
the sagging ASP market, but there’s optimism that IP storage introduces several
promising applications for outsourcing: network client access to remote
storage, remote mirroring between storage controllers, remote backup/recovery,
business continuation, resource consolidation at a secure hosting center are
all viable opportunities.
Tempering the Enthusiasm
The promise is evident, but the path forward has some (ahem)
minor obstructions. The Internet (IETF) standards for transporting block
storage over IP aren’t ready yet, so interoperability problems may initially
deter or plague companies who place a heavy emphasis on this requirement. Conventions
for resource naming and discovery mechanisms are needed. Storage applications
often require bounded latency, so IP deployments may require DiffServ or MPLS
support for quality of service assurances.
Once IP addressable, storage devices and networks must
incorporate security measures sufficient to defend against a panoply of
threats. While strong authentication, data origin verification, and data
integrity and confidentiality are under consideration by the standards
community, it’s essential that organizations factor storage security into their
security policy and implementation.
Consider these carefully as you select product, and be
certain that your vendor of choice has a well-conceived path towards a
standards based solution. But don’t hesitate to consider networked storage. Now
that it’s IP-based, it’s very likely to factor into your future networking.
Return to CoreCom's Technology Corner
|