One of the
more interesting dynamics regarding storage these days is that as
the volume of data being processed by organisations across the
board continues to mount, the price of storage continues to
fall.
As a consequence,
even through the more traditional storage options are increasingly
incapable of satisfying demand, the new offerings are ore suited to
the needs of more business irrespective of their size.
So of what do
these choices consist and how do they differ from more traditional
storage models?
The first and most
expensive of these new technologies is the storage area network
(SAN), which according to experts, comprises an entire network
dedicated to storage and is best suited to organisations running
large databases.
“It works best
where companies want very high performance and to really manage,
their data resources deeply because it addresses data at the block
level. The problem is that the investment in SAN technology is
quite significant. It’s not got a particularly low entry point and
you also have to buy a lot of new infrastructure,” explains John
Hammond, commercial director at value added reseller, Logicalis
Computing
Traditionally, SAN
devices have been connected using the SCSI protocol over fibre
channel, a gigabit speed networking technology, which, in cost
terms, is out of the immediate reach of many companies, especially
smaller firms.
Yet initial
investment is but a small part of the overall issues that need to
be taken into consideration, argues believes Sue Clarke, a senior
analyst at Butler Group. “SANs are not simple systems. They’re
complex and organisations will need third party help to set up at
least some of it and then they may face a lack of internal resource
to operate and manage it,” she explains.
A general shortage
in IT personnel trained in the relevant SAN–related skills makes
using SANs a non-trivial issue for the largest of companies; a
situation exacerbated in those organisations without an IT
department.
While this
situation undoubtedly presents an opportunity for suppliers to
provide value added services, it can also lead to difficulties in
your buying process. That is to say, you may need to convince your
finance directors that investing in a fairly significant capital
infrastructure project is worthwhile.
“If you put in a
new application, you can often say that the return on investment is
this and you’ll get your money back in x number of months. But it
can be more difficult to justify the expense of networked storage
and especially more expensive SAN technology,” outlines Clarke.
“The finance director is likely to say ‘our current storage is
working now and disks are cheap so why don’t you just put another
one in’. This means it’s necessary to put the case in terms that
the business will understand such as networked storage will make it
easier to do disaster recovery.”
The second
category of networked storage technology, meanwhile, is
network-attached storage (NAS). Unlike SANs, which comprise
separate specialised networks, NAS consists of storage devices that
can be plugged into a general-purpose IP-based local area
network.
As a result, NAS
systems are cheaper to set up and run than SAN-based ones, but, on
the downside, are also slower. They are, however, quicker than more
traditional direct-attached storage (DAS) and are also simpler to
manage because the hard disks of which they comprise can be looked
after from a central point.
Other advantages
include higher levels of scalability, reliability and availability
than DAS, which comprises a single storage device such as a tape
drive, SCSI hard drive or on-board RAID array that is connected to
an individual server.
A key problem with
the DAS approach, however, is that, not only does it use a lot of
CPU processing power, but each device has to be administered,
maintained and backed up separately, which can cause you management
headaches, particularly when large volumes of data are
involved.
“Many
organisations are starting to move away from DAS because of its
lack of manageability as the volumes of data they have to manage
keep growing. It’s about the complexity of managing that
environment so we’re tending to see more consolidation, with people
moving to bigger servers and networked storage alongside that,”
says Clarke.
It would be unwise
to infer though that DAS has had its day and is essentially
yesterday’s technology. Explains Clarke “For example, you might not
want an application handling sensitive financial information to be
available on the wider network. So there’ll always be a need for
some degree of DAS and small organisations are also unlikely to
have the volumes of data necessary to justify a move to any form of
networked storage.”
Similarly, NAS was
considered by some as “the poor man’s SAN” when it first came out
and regarded by the same, somewhat simplistically, as nothing more
than something that bridged yesterday’s technology in DAS, and
tomorrow’s in SAN. It is evident that NAS is now coming into its
own, gaining increased recognition for its strengths of file- and
data-sharing, areas in which SANs are weak.
Hammond explains:
“NAS doesn’t have the throughput rates needed for database
applications, but it works best with more basic file storage
applications that aren’t database-hungry such as email and file and
print.”
Nonetheless, the
technology is also ideal for those businesses that cannot afford
big upfront investments. As Robin Burke, a vice president of
research at Gartner, points out: “Anyone can plug and play an extra
500 GB or even a terabyte into the network and you don’t need
dedicated IT staff to do it.”
This can lead to a
number of business benefits. Now that the differences between the
two networked storage technologies are now becoming less marked, it
is becoming increasingly possible to get the best of both
worlds.
Graham
Titterington, a principle analyst at Ovum explains: “Networked
storage is not a new concept and it’s been around for a few years
now, but what is new is that the dividing line between SAN and NAS
is getting more blurred. In the past, the two technologies were
quite separate, but they’re now coming closer together.”
For example, the
increased availability of SANs based on iSCSI, an IP version of the
SCSI networking protocol, means that they are becoming cheaper to
implement as organisations can use IP rather than expensive fibre
channel networks to connect their storage devices together.
While adoption of
such technology is still in the early stages, into the future, it
is likely to prove a popular choice. “A SAN-over-IP is now an
option for SMEs that could never have thought of introducing one if
they had to put in fibre channel,” says Clarke. And there are added
benefits in terms of staff, an issue as stated, has plagued SAN
usage. “The use of IP networks in SANs is quite a new phenomenon.
It’s largely about economy and ease of use, not least because it’s
easier to find people with IP skills,” says Titterington.
But iSCSI has also
begun to appear in the NAS world. The pioneer in this space was
Network Appliance, which introduced both fibre channel and iSCSI
support into its NAS devices to give them SAN-like qualities,
although other vendors have since followed suit.
As to which types
of organisation are using network storage technology, in all of its
manifestations, however, this is mainly driven by the amount of
stored data they are trying to tackle.
“As a rule of
thumb, the bigger the organisation, the more likely adoption is,
with SANs generally deployed in the data centre and NAS in
mid-sized companies or at the departmental level. But that’s not
the whole picture,” says Titterington.
For example, a
media or publishing company generating huge amounts of data but
employing only a small number of staff is more likely to introduce
networked storage than a larger company doing manufacturing. The
same applies to highly regulated industries such as financial
services that may well be penalised if they don’t archive
information.
“Another reason
for deploying networked storage is to improve the backing up of
data for disaster recovery and business continuity purposes. If
you’ve got a lot of computers operating separately and you need to
back them up every few minutes to ensure they’re up-to-date, it’s a
real pain and very time-consuming to have to do it for each one
individually,” Titterington says.
The bottom line
for many organisations, however, is simply trying to keep on top of
the 50% year-on-year increase in data needing to be stored and
retrieved, and deploying the most suitable technology available to
manage the complexity involved in doing so.
Despite this says
Nigel Lambert, channel director at networking and storage
distributor Zycko, take-up in by smaller concerns is still only at
a fairly early stage, although he is seeing demand growing
rapidly.
“If you take the
top 5% of enterprises, adoption is pretty mature, but the sweet
spot of the market is mid-sized companies and medium-to-large
enterprises. While demand is across the board, network storage is a
must-have now in local government, financial and insurance and
professional services firms due to compliance issues,” he says.
Vendors complain
that the main barrier to more widespread adoption at the moment is
a lack of understanding of what networked storage is and what
benefits it can bring. “There are a lot of network storage
technologies and the technology itself is complex, which is very
confusing for people. So one of our key jobs in the channel is to
remove that confusion and help customers to move forward,” believes
Lambert.
Yet among some
dealers, confusion also exists. “Until recently, the networking and
the storage channels have been very separate, with the IP
community, for example, knowing very little about storage and the
storage community knowing very little about IP. That is starting to
change, however, with data networking channel in particular,
starting to get more involved in the storage world,” Lambert
says.
In the future, we
will likely see the advent of technologies such as storage
virtualisation, which involves putting a software layer over
different physical storage assets to provide a single logical view,
which makes them easier to manage.
The software also
improves resource utilisation and eliminates vendor lock-in as it
can run on any hardware, which means that different devices can be
swapped out and replaced with minimum disruption.
And while this and
other storage management software may still be relatively mature
and at an early stage of adoption, it is likely to become
progressively more valuable as storage complexity continues to grow
in all sectors and for all organisations.