With every server purchase comes a basic question: How much
storage should you include with the new server purchase? For larger companies,
you might have already addressed this question through the implementation of a
storage area network. However, until fairly recently, centralized storage
pricing has remained out of reach for many small- and medium- size
organizations. Regardless of the specific situation at your company, one thing
is certain: there are a lot of choices – and a lot of decisions to be made – in
today’s storage marketplace. In this article, I’ll provide an overview of the
current technology as well as its pros and cons. Use this article in
conjunction with the Storage product
comparison worksheet
available for download from TechRepublic to get a
better idea about the various options available for your business. In a future
article, I’ll provide a more in-depth view of each of these solutions.

Storage options

There are three overall storage options that deserve
consideration: direct-attached storage (DAS), network attached storage (NAS),
and a storage area network (SAN). As you might expect, each option meets
specific needs and has pros and cons that you must weigh before making a
decision.

Each storage option is discussed in detail below.

Direct-attached storage

Just about anyone who’s ever touched a server is familiar
with DAS. DAS is the storage that fits right into the server or that hangs off
the server. For example, storage connected to a server’s external SCSI channel
is also considered direct-attached storage.

DAS has enjoyed a long history and is still the storage of
choice in many situations. With the fastest transfer rates between the disk
system and the server, DAS is ideal in situations where fast disk access is
demanded, although some new SAN equipment is starting to give DAS a run for its
money in this department. Further, most applications have no trouble at all
working in a DAS environment, so you don’t usually need to worry about
application issues and can focus on other areas that might cause problems
instead.

Not all is rosy with DAS, however. First and foremost, IT
managers must constantly contend with “space issues” which attend to
answer to common questions:

  • How
    much storage do I need to provision for a new server?
  • What
    do I do if my provisioning wasn’t quite right and I need to add space?

Some options on the market help to alleviate the storage
burden related to these questions, but either way, you still need to make
best-guess estimates regarding storage and then expand that storage as unanticipated
needs surface.

Second, you still have to manage most DAS on a
server-by-server basis meaning that you need a mechanism in place to monitor
server disk usage per physical unit. Not many IT managers relish the thought of
running out of disk space in the middle of the day!

DAS is ideal in a number of situations:

  • When fast access to storage is
    required, but the newest SAN technology prices remain out of reach or are
    not necessary.
  • For very cost-conscious customers,
    DAS will remain the least expensive storage mechanism for a long time. Of
    course, this is only in terms of hard physical media costs. When a full
    comparison with other technologies is completed that takes into
    consideration administrative overhead and storage efficiencies, you might
    find that DAS is not at the top of the chart anymore.
  • For very small environments that
    just don’t need anything more.

Network-attached storage

There are some times that you just need to throw storage on
the network, accessible by a number of users, and call it a day. Enter
network-attached storage (NAS). NAS installation is usually very simple. Like
DAS, though, you need to answer some basic questions about how much storage you
need for your particular task. Unlike DAS, NAS devices can usually be more
easily expanded in the event you need additional capacity. For example, where a
commonly available DAS device tops on in the 2-TB range, some NAS devices on
the market can scale upwards of 200 TB. With a few exceptions, a NAS unit is a
perfect device for situations where you just need to throw storage at a
problem.

There are a couple of major exceptions to this rule of
thumb: Most importantly, situations in which block-level access to data is
required, such as for databases and Exchange information stores, are not
appropriate for NAS units. Some NAS units do support these processes, but it’s
a case of forcing a tool into being appropriate for a job. Second, when you
really need high-speed access to storage from a server, a NAS may not be
appropriate since all data needs to traverse the network and is thus limited by
the speed of your network.

There are two situations in which NAS devices really shine: first
and foremost, Web serving, and in a very close second, general file storage. Both
applications require significant disk space, but direct data access from a
server is seldom required. Instead, since most data from these two types of
stores is accessed over the network anyway, it doesn’t matter that it’s coming
from a NAS vs. DAS hardware.

There’s also another reason that NAS devices are really good
for Web and file serving, but not for databases and Exchange stores. It has to
do with what’s called file-level vs. block-level data access. In file-level
systems, data is accessed by a file name, as the name implies. In block-level
systems, data is accessed using a block address, which is the location in which
specific data is stored. In a client/server scenario, when you request a file from
a file server, you’re asking for a specific file and the server does the block
read to get that data for you. Databases and Exchange stores have difficulty
communicating in this way, so their stores are not appropriately stored on a
NAS device. Databases and Exchange stores are more efficiently accessed using
block-level functions available with DAS and SAN solutions.

Even though NAS is a great solution when you need to throw
storage at a problem, it does have some drawbacks:

  • It can be more expensive than
    similar DAS space.
  • Not appropriate for some
    high-usage tasks like databases and Exchange stores.
  • Data retrieval is only as fast as
    the network connecting the unit.
  • A potential single point of
    failure in the storage infrastructure.

Storage area networks

The grand pooh-bah of storage, the SAN is the most expensive
storage option of the bunch, as well as the most complex. However, SANs provide
capabilities not found in other solutions and, in the right situation, can
actually end up saving a company some funds, even considering the expensive
initial outlay.

SANs today come in two flavors: fibre channel, and iSCSI or
IP-based SANs. Fibre channel is the most well known type of SAN, but over the
last couple of years, iSCSI-based SANs have started to hit the market in a big
way, mainly due to their good performance and much lower cost versus fibre
channel.

SANs truly combine the best of both NAS and DAS storage. For
example, with a proper implementation, you get a completely redundant storage
network that is eminently expandable to, literally, hundreds of terabytes a la
NAS, but you also get block-level access to the data just as you get with DAS. You
also get access to data at a reasonable speed, making SANs good even for
operations that require significant disk access. With a SAN, you also get
centrally managed storage with the ability to provision space on-the-fly. Even
better, with some implementations, you can even configure your servers with no internal storage and require that all
systems boot directly from the SAN (fibre channel only). Talk about plug and
play!

With all of these great points, what are the downsides of a
SAN? There are two major drawbacks to a SAN: cost and complexity, particularly
when it comes to fibre channel implementations. A reasonable fibre channel SAN
can start in the $50-60K range for
just a terabyte or two of storage. On the other hand, newer SANs based on iSCSI
might start in the $20-30K range, but aren’t quite up to the performance levels
of their fibre channel cousins. The difference in price is mostly due to
iSCSI’s ability to make use of off-the-shelf gigabit Ethernet hardware, whereas
fibre channel requires specialized, expensive equipment.

The chart

When making a storage decision, it’s important to have all
available information at your fingertips. This article was meant to be a primer
on the different types of storage and help you to move in the right direction
regarding your environment. To sum things up, I’ve provided the quick-glance
chart below to help you compare and contrast the different kinds of storage. The
SAN category is broken up into iSCSI and fibre channel to help you make the
distinction between the two technologies.

 

DAS

NAS

iSCSI/IP
SANs

Fibre
channel

Price

Low

Medium

Medium-High

High

Expansion

Limited

Depends on solution

Depends on solution

Depends on solution

Management

Inefficient

Inefficient

Very efficient

Very efficient

Fault
tolerance

Somewhat tolerant

Somewhat tolerant

Very tolerant

Very tolerant

Good
for file storage

Yes

Yes

Yes

Yes

Good
for database storage

Yes

No

Usually

Yes

Good
for Web serving

Yes

Yes

Yes

Yes

Good
for Exchange stores

Yes

No

Usually

Yes

Ease
of installation

Simple

Simply

Somewhat difficult

Very difficult

Disaster
recovery capabilities

None

None

Many

Many

OS
support

All

N/A

Windows, Linux, UNIX, NetWare (others
dependent on drivers)

Windows, Linux, UNIX, NetWare (others
dependent on drivers)

Primary
vendors

Any server vendor

IBM, Dell, HP, Network Appliance

LeftHand, EMC, HP, IBM, Network
Appliance

IBM, EMC, HP, Network Appliance