Organizations are clamoring for better ways to serve up files to
meet the insatiable demands of applications and network users.
Traditional file server computers are far too small for many modern
businesses, and large
storage area networks (SAN) are often too
expensive and complex.
Network attached storage (NAS) technology
has filled this gap, providing powerful file serving capability
with huge storage capacity and an increasingly comprehensive
suite of enterprise-class features. NAS has emerged as a primary
storage resource in the small and midsized business (SMB) and
has proven itself in the enterprise. "I would say that every one
of those [Fortune 100 or Fortune 200] companies has some form of
NAS file sharing -- whether it's an appliance or Windows file
serving in their organization," says Greg Schulz, founder and
senior analyst at the StorageIO Group in Stillwater, Minn. This
article covers the essential concepts of NAS, highlights key
implementation issues, shares some real-life NAS case studies
and looks ahead at the future of this technology.
Understanding NAS
In its simplest form, NAS is a disk-based storage system with an
IP network interface that is designed expressly to serve files on
an IP network. This dedicated file serving offers vastly greater
storage capacity than traditional file server computers that
typically hold just a few disks. NAS is also noted for its superior
connectivity and storage performance, allowing applications and
users to be serviced faster and more reliably than common file
servers can handle. NAS is recognized for three principal benefits:
consolidation, deployment simplicity and ease of management.
Before NAS, there might have been hundreds or even thousands of
individual file server computers scattered throughout an enterprise
-- each with a few disks, and each demanding maintenance or
configuration from IT staff. Today, a NAS system can replace a huge
array of file servers, @32605 consolidating terabytes (TB) of
storage into a single system at one location. A NAS is also simple
to deploy, usually requiring little more than a power source and an
IP network connection for basic operation. A proprietary operating
system is frequently included to enhance performance and ensure
security.
RAID is a common feature for data
protection, and some NAS systems now support dual-parity RAID
(RAID-6 or RAID-DP) to guard low-cost
SATA storage against multiple simultaneous
disk failures. More disks can easily be installed to add storage
capacity, and if the NAS box fills to capacity, a new NAS box
can be purchased and connected with equal ease. Finally, the
consolidation and management of NAS resources can typically be
managed through a single management software tool. Often, the
management software can support multiple heterogeneous storage
systems, easing management overhead even further.
NAS has evolved along with
Fibre Channel SANs, and the two technologies
appear to have divided into separate storage camps. NAS has
typically taken on the role of low-end storage, while Fibre
Channel SANs have been associated with high-end performance.
While this may have been a justifiable segregation in years
past, industry experts say that faster Ethernet and better NAS
system designs are changing this relationship. "Historically,
there's been some cost advantages to NAS, which has
inadvertently pigeonholed NAS to be perceived as cheap 'toy'
storage -- not befitting an enterprise," Schulz says, noting
that contemporary NAS systems can support the storage tiering,
high performance and high availability that had only been
available in SANs.
Although we often refer to NAS as a dedicated appliance (e.g., a
NAS box) with its own internal storage, some organizations may
choose to forego a new appliance when a SAN already exists in the
enterprise. Rather than a NAS appliance, a NAS gateway can be used
to serve files from SAN or some other external storage. The
presence of a gateway adds NAS functionality, but uses the SAN or
separately attached disk array for actual storage. The advantage of
a gateway is that NAS storage is not limited by the capacity of the
appliance itself, but rather by the storage capacity of the
separate array or SAN, which can be considerably larger. For more
details on NAS appliances and gateways, see the
NAS Upgrades Buying Guide.
Today, experts agree that NAS has earned a permanent place in the
enterprise alongside the SAN. SANs remain the preferred choice for
block-based storage, but NAS has become a staple of file storage,
especially in its ability to serve unstructured file data, which is
growing rapidly in the enterprise. NAS can easily exist alone, and
many smaller organizations that are moving beyond file servers will
adopt NAS. But SAN and NAS technologies can also coexist when the
storage environment handles a mix of block and file data. "In a
large environment, every good SAN should have a NAS -- every good
NAS should have a SAN," Schulz says.
Implementing NAS
NAS implementation typically involves the deployment of an
appliance or gateway device, along with the
configuration/management software that accompanies it. Experts say
that most NAS deployments are turnkey, requiring only simple setups
for basic operation. High-end NAS systems can be a bit more complex
and generally involve more sophisticated tools, such as backup,
replication and snapshot software. Some software may require a
separate purchase, so it's important to note what functionality is
available out of the box, and which features may require a separate
purchase.
Keep an eye on the NAS network traffic. Without adequate
bandwidth, a network segment hosting the NAS can easily become
overloaded and impair performance. In most cases, a NAS is
implemented with multiple
Gigabit Ethernet ports that can be
aggregated for improved bandwidth. Multiple ports also allow for
failover, an essential element of high-availability operation.
Performance is also a serious issue when the NAS is called upon to
handle transaction-centric applications, such as databases or email
systems like Microsoft Exchange. SANs are still the preferred
choice for storage in heavy transactional environments, but
high-performance NAS systems are offering alternatives when the
application workload can be balanced with the NAS system. "You
start to run into limits with NAS at the very high end where you're
serving up hundreds or thousands of database transactions per
second," says Phil Goodwin, president of Diogenes Analytical
Laboratories Inc., noting that very intensive
OLTP environments can have trouble being
hosted on NAS.
As an added wrinkle, a NAS can sometimes influence software
support. For example, some block-based applications may not even
support deployment on file-based storage, so be sure to review the
applications that you intend to host on NAS and verify that the
application vendor will continue to support those applications. Lab
testing is an excellent way to see the application's behavior and
stability on a NAS prior to a major rollout.
The actual effort needed to manage a NAS is typically not
significant, and good tools can ease the overhead. Opt for tools
that will support multiple NAS systems and heterogeneous storage
systems wherever possible. This will prevent the IT staff from
having to deal with multiple management tools as more NAS systems
are added. Otherwise, the ease of NAS deployment can eventually
result in a myriad of NAS systems and file servers -- the sheer
number of NAS systems can ultimately cause management headaches.
"In the PC and server world, the solution is some sort of
consolidation or virtualization," Schulz says, "and we're going to
start seeing more of that [virtualization] in the NAS space."
The impact of NAS
NAS usually helps to overcome the expense of a SAN or the
management headaches of DAS. For Jeremy Whaley, director of
information systems and network services at Claremont McKenna
College in Claremont, Calif., NAS overcame both issues. Several
years ago, the college employed an IBM-based SAN with 4.5 TB, with
another 3 TB spread out between various DAS systems connected to
servers as needed. The expense of server HBAs left the SAN poorly
utilized. "Only a handful of Windows file servers were taking
advantage of the SAN," Whaley says. SAN performance had also proved
to be problematic, and frequent SAN reboots interrupted service to
users across the campus.
For Whaley, the challenge was to improve storage performance while
keeping costs in line and begin a process of server consolidation
that would support more storage on fewer platforms --- while
eliminating the security and patch headaches with a proliferation
of Windows file servers. The college opted for the simplicity of IP
network connectivity. "We knew we needed to fix, replace or
redesign the storage facilities for the future," he says. "The
reason that we landed on NAS rather than another SAN is because we
wanted to actually retire and consolidate our file servers -- of
which there were many." After about six months of consideration,
the college settled on a BlueArc Corp. Titan NAS system. Whaley
notes that BlueArc also supports iSCSI, so there is always a
potential to add an iSCSI SAN later if necessary. See the article
ISCSI Explained for more information.
After installation, SAN data was migrated to the NAS using the
Robocopy utility. Testing and final activation took place at the
end of December 2006. There were numerous technical issues to
resolve with network segmentation, but Whaley reports a
problem-free deployment of the NAS itself. Today, the NAS provides
19 TB of usable storage to the college. Retiring individual file
servers has considerably reduced server management and maintenance
issues, eased network congestion problems, and vastly improved
storage reliability. "After we cut over [to the NAS] we continued
monitoring the two side by side, and the old SAN has already had
three errors where we would have had to schedule maintenance in the
evening with IT staff on campus," he says. He also noties that
remaining file servers have been very stable.
More storage will undoubtedly be added to the NAS system into the
future, but Whaley is also considering a move to virtualize
remaining file servers -- further consolidating the infrastructure
and management overhead. Beyond that is the prospect of disaster
recovery for the college. "This particular NAS device has a really
wonderful replication feature built in," Whaley says. "If you have
a second unit out there, you can replicate between the two in
several different fashions."
Healthcare providers have seen a dramatic increase in their raw
storage to meet the demands of digital medical images. The Capital
Region Orthopedic Group in Albany, N.Y., struggled with outdated
practice management and surgery systems offering about 200 GB of
usable storage. With the move to a new facility back in 2000, the
adoption of an updated practice management system and a shift from
conventional film to digital X-rays in 2003, the group's storage
demands exploded. "We went from a mom-and-pop shop to an enterprise
almost overnight," says the Group's chief technology officer,
Raymond DeCrescente, Jr. The incessant need for storage drove the
adoption of a Hewlett-Packard Co. (HP) SAN. An HP b3000 NAS head
accesses a current total of 8 TB SAN storage for the group's
updated picture archiving and communication system (PACS). Each
digital X-ray study requires 25 MB. "We basically are filling our
storage now at 5 GB per day," he says.
The choice of vendor came down to a matter of convenience and
capability. DeCrescente notes that access to local HP engineers and
their ability to redesign a data center from the ground up were
critical factors. This single-source approach avoided the need to
tackle numerous storage and infrastructure upgrades through
separate vendors. DeCrescente is particularly pleased with the
scalability of his storage, noting that current 144 GB drives can
be upgraded to 300 GB Fibre Channel drives. This essentially
doubles the storage to 25 TB. By installing additional drive
shelves, the SAN can support over 50 TB. "I'm planning on doing
just that, and then I can do
disk-to-disk [D2D] snapshots," he says.
It took about a month to test and burn in the new infrastructure,
but DeCrescente is pleased with the reliability and call-home
capability of the storage. Although he cannot quantify the
return on investment (ROI) or other
financial impact of his storage choices, the benefits are clear.
"You can't help but to save time on the maintenance." The one
bit of advice DeCrescente offers is to take full advantage of
training opportunities. "If I look back on it, I would have done
that training during that month while everything was burning
in."
After DeCrescente adds storage, he intends to add HP data
protector and snapshot software for enterprise-class data
protection. A hardware refresh is planned in another several years,
which will move the current SAN and servers offsite and install a
new HP SAN in the main office -- allowing for D2D disaster
recovery.
The future of NAS
Experts agree that NAS has a bright future in all types and sizes
of enterprise, noting that NAS technology is evolving to meet more
demanding tasks. Traditionally, NAS is deployed to support
file-based applications in light-to-moderate traffic storage
environments. But today, NAS use is expanding down into the small
office/home office (SOHO) and up into the enterprise, and the lines
between block and file storage are blurring. More block
applications are being deployed on NAS systems where they would
have appeared on SANs in the past. NAS systems are also
increasingly capable of handling database and other
traffic-intensive applications. Such improvements are often matched
with better management tools and high-end features, like
replication and snapshots. "There are some very large companies
that run almost exclusively on NAS," Goodwin says. "I think as it
becomes more capable, as people deploy more gateways and provide
virtualization to their environment, NAS will continue to expand
its presence."