IT directors looking for an easy way to manage
increasing volumes of data by buying more hardware could be storing
up problems for the future.
When it comes to tackling burgeoning storage volumes, most IT
directors admit they tend to choose a short-term solution instead
of a long-term strategy. In fact, 80% say they simply purchase more
hardware rather than investigate ways of getting more out of
existing resources, according to a recent survey conducted by
market research company Vanson Bourne.
In the survey, commissioned by systems integration company
Computer Systems and Finance, many IT directors admitted that
plummeting hardware costs made buying extra storage seem like a
"relatively pain-free" option.
But in addition to storing corporate data, they may also be
storing up long-term problems for themselves, says Mark Sweeney,
business development director at Computer Systems and Finance. "The
policy of adding more storage to an infrastructure that is not
designed to support it can create its own problems,” he says. “Such
organic growth quickly becomes costly to maintain and manage –
especially if the hardware has been bought from different suppliers
to capitalise on tactical cost savings."
Such an approach can introduce much inefficiency. Storage
management and administration software tools may be duplicated
across the infrastructure; storage administrators have to retain a
wider base of product knowledge; and parts have to be bought from
separate suppliers so the company has to forgo volume discounts
from individual suppliers.
And there is another problem: although organisations continue to
purchase additional storage hardware in response to tactical
business demands, it is likely that much existing storage space is
sitting idle. "It is accepted throughout the data storage industry
that many organisations use only 40% of their storage capacity,"
Sweeney says.
What is needed is a new approach in which IT managers set out a
long-term strategy for their storage architecture. To do so, they
must take the time to evaluate existing levels of use and the
workload their storage infrastructure is forced to bear.
The larger IT companies say they help their customers to do
this. "It is vital to analyse current storage usage before
embarking on any storage implementation or upgrade,” says Nigel
Ghent, UK marketing director at storage supplier EMC. “You need to
know what you are holding, where you hold it and what data has
specific compliance requirements. That is a huge job in
itself.”
Steve Cliff, sales manager for the northern region at IBM, says
a company should then look to implement storage strategies such as
information lifecycle management. "Once you know what data you need
to store, you can put in place a variable cost storage architecture
where you match storage systems to data, according to its
value.”
The next step along this path is to network the storage
architecture. In recent years, the evolutionary theme in data
storage has been the shift away from direct-attached storage (DAS)
to storage area networks (Sans) and network-attached storage (NAS)
systems. Although DAS still persists in most datacentres, and will
continue to do so for as long as the legacy applications it
supports remain in service, new storage investment has largely gone
to building San and NAS systems.
"Traditional storage topology - pairing a storage array with a
single server, or statically partitioning an array to meet capacity
needs for several servers - fails to provide the flexibility to
dynamically reallocate resources to meet unpredictable demands,
isolates data to a single server and, given data storage growth
rates, is almost impossible to manage," says Bob Zimmerman of IT
market research company Forrester Research.
Networked storage, by contrast, circumvents this problem, in
theory at least, by organising individual storage resources into a
unified "pool" that can be accessed by a variety of applications
across the company.
This also enables companies to implement storage systems from
different suppliers, thanks to standardisation efforts in this area
(see “Raising The SMI-S Standard” below). "With a good, solid San
in place, I think it is safe to say that most organisations can
comfortably continue to invest in storage systems from a number of
different suppliers," says Martin Davies, principal consultant at
systems integration company Morse.
Most organisations have implemented a blend of San and NAS
technology. San systems provide the fined-grained, so-called
block-level access to data, which is needed to optimise the
performance of databases and database-centric applications such as
customer relationship management and enterprise resource
management. NAS, on the other hand, better supports the kind of
file-level access needed for document- or file-centric applications
such as office, workflow and collaborative environments.
Thanks to San-NAS convergence, all data can be held on a single
physical San, but the system's capacity can be logically
partitioned to accommodate block-level access over Fibre Channel,
SCSI or iSCSI, and file-level access over IP.
Users must also choose an appropriate communications protocol
based on price/performance parameters. Fibre Channel still offers
the best price/performance balance, and can operate across
distances of about 100km. To link more distant datacentres, Fibre
Channel can be encapsulated in IP packets, and where global reach
is needed, users can deploy Fibre Channel over wave division
multiplex links.
The drawback of Fibre Channel is cost. Fibre Channel host bus
adaptors are expensive and, where their expense is not justified by
service level requirements, slower, shorter distance links can be
achieved using commodity disc interfaces such as serial advanced
technology attachment (SATA), or the economy-class technology of
SCSI over IP (iSCSI).
iSCSI is increasingly spoken of as a cheaper, mainstream
alternative to Fibre Channel. However, most tape and disc devices
still do not have native iSCSI connectivity, so any perceived cost
saving will normally be cancelled out by the need to install an
expensive bridging device.
SATA, a more robust version of the ATA interface built into
commodity, PC-class discs, offers a cheaper, near-line alternative
to Fibre Channel disc. Back-ups can be sent from SATA to tape
without disruption to the production environment.
Similar price/performance issues come into play when archiving
to tape. DAT, VX1 or AIT-1 are all popular, economic formats for
backing up local servers, but they do not offer the performance
required for larger systems. Here, DLT (digital linear tape)
devices supporting transfer rates of 2.5Mbps are a better option
and are moderately priced.
However, DLT has reached the physical limit of its capacity
(80Gbytes), forcing larger users to invest in high-speed,
high-capacity storage such as SuperDLT, LTO and SAIT. Using 2:1
compression, these formats can handle up to 600Gbytes, 400Gbytes
and 1.3Tbytes of data respectively.
Managing storage resources effectively is also a software issue.
Storage resource management software tools aim to automate much of
the labour-intensive and error-prone processes associated with
allocating storage resources, and virtualisation enables the
servers and their applications to treat the devices on the storage
network as a single pool of capacity.
By allowing more efficient and flexible use of storage resources
with less manual involvement, virtualisation and storage resource
management promise improvements in IT staff productivity.
Given the complex questions that surround the purchase and
implementation of company-wide storage architecture, it is hardly
surprising that some IT directors prefer to turn storage management
over to a third-party services company. Despite that, the storage
service provider model that was touted three or four years ago by a
number of companies heavily backed by venture capital failed to
take off, for a number of reasons.
Many customers did not want to give up control of their data and
the customers that did wanted dedicated resources, thus negating
any potential benefits that storage service providers got from the
economies of scale offered by sharing resources between clients.
The storage management tools that enable IT directors to allocate
storage resources to data were not available at that time so
storage service providers were forced to write their own.
But most of all, many clients raised the question, "Why
outsource storage in isolation?" In other words, if they decide to
outsource storage, it made more sense to outsource servers as well.
For that reason, storage is now usually outsourced along with
servers and other infrastructure elements as part of a wider
managed services deal.
For example, in November 2004 French electrical group Schneider
Electric signed a £1bn 10-year outsourcing deal under which IT
services company Capgemini would take responsibility for the
support of Schneider's entire IT infrastructure - including storage
systems.
There is, however, a booming market for specific storage
services such as back-up and recovery and remote mirroring. In
mid-2004, broadcaster GMTV signed a three-year contract with
hosting company PSINet Europe to manage its disaster recovery and
data back-up systems. GMTV's production and broadcasting systems
are archived and stored off-site to ensure a programme can go ahead
even if the broadcaster's London studios are evacuated.
These kinds of services may be attractive to some companies, but
Vanson Bourne's survey found that, despite ever increasing volumes
of corporate data and a slew of legislation about their retention,
40% of IT directors still prefer to keep storage in-house. To do
so, they must plan for the long term, not look for a quick fix.
Raising the SMI-S standard
In 2004 the major players in the storage industry finally agreed
on a standard application programming interface that allows storage
management software to link to any disc array, from any supplier.
The Storage Management Initiative-Specification (SMI-S) may have
limitations in its current version (1.01), but promises true
multisupplier interoperability in the future.
Five years in the making, SMI-S has come on "in leaps and bounds
over the past 12 to 18 months", says EMC’s Ghent. "Many people do
not understand the full extent of the work that has gone on behind
the scenes - but a standard must have teeth, and SMI-S does."
SMI-S is based on the Distributed Management Task Force's common
information model and web-based enterprise management standards. In
the storage world, SMI-S is similar to the simple network
management protocol, the standard for managing and monitoring
network devices on a Lan or Wan. More than 100 products - from
suppliers such as EMC, Hitachi, IBM, Hewlett-Packard, McData,
Brocade, Network Appliance, StorageTek and Sun - are certified as
SMI-S-compliant, and the number is growing.
That broad support for SMI-S is vital because it guarantees IT
decision-makers a common schema for choosing storage management
applications and hardware. An IT director can order a
Hewlett-Packard disc array now and an EMC SMI-S-compliant disc
array in 18 months’ time in full confidence that the same software
can handle such tasks as the provisioning and monitoring of
both.
SMI-S 1.01 provides common schema for the provisioning of a
storage network, the monitoring and provisioning of storage
subsystems, and for creating, mapping and masking logical unit
numbers - a unique identifier used on a SCSI bus that enables it to
differentiate between up to eight separate devices. Key features
still to be addressed are snapshot management, replication,
mirroring and support for NAS devices and tape libraries. These are
expected in SMI-S 1.1, due in 2005.