Till now, the approach has been to invest in servers on which the
applications run. These servers are in turn attached to storage devices.
However, this approach creates islands of data storage. The number of servers
and applications running over them would reflect the number of storage devices.
Managing a scattered DAS is a nightmare for any CIO.
"Multiple servers and scattered data created a complex storage
environment. How to get maximum value from this network has led the
organizations to rethink their network strategy," said Manoj Chugh,
president, EMC India and SAARC.
The mindset has changed to consolidate applications and servers on a single
intelligent infrastructure.
Storage Planning-When, Where, How?
The storage industry has been trying hard to move away from the scattered
environment and creating a single pool of data, which could be accessed by all.
Instead of spreading horizontally, effort has been made to have multiple layers
of data storage based on the behavior and importance of data.
"There are four broad areas-manageability, availability, performance,
and scalability-according to which a CIO plans his storage," said Shankar
Subramaniam, sales engineer manager, Southeast Asia-Pacific and Korea, Brocade
Communication.
"There is no single storage design that can fit every organization.
Designs and solutions vary according to the requirements," he added.
The first step towards a storage network is to quantify data and its growth
rate. Once the amount of data on a network has been assessed, the CIO has to
identify how much information needs to be shared and by whom. The information is
then broken down into various categories according to its importance. This gives
the CIO an idea of the online data at any point in time.
Designs are planned based on data growth and usage, the number of ports, port
connects, device attachments, and connectivity. The initial evaluation of growth
data and the business applications being deployed also determines the various
levels and management of storage.
Data classification is critical to decide on the type of storage
architecture. Tiered storage gives you the flexibility to manage your data
better and scalability issues are also taken care of.
The tiered storage normally has a combination of fiber channel and ATA
drives. Then there are specific devices for archival purposes. Normally, tapes
are used as the final resting places for information. As more emphasis is laid
on the management part, vendors are incorporating intelligent software to move
data across the storage tiers.
"Information management is very important as it optimizes the storage
space. Only essential information is kept online and rest is pushed back
depending on the priority being attached to it," Chugh said.
Advantage SAN
The storage is increasingly moving from DAS to network area storage (NAS).
And within NAS, storage area networks (SAN) are getting a large chunk of the
market share.
"There is a common belief that SANs are complex and expensive. Contrary
to this, SANs enable better management of data and better resource
utilization," said Leong Kam Hong, product mktg. manager, Dell/EMC storage.
SAN works on shared storage space and hence provides access to multiple users
at the same time. As the number of servers decrease, the manpower required to
maintain them also reduces. And the intelligent software sitting on top of the
SAN appropriates storage space management.
"Though implementing the first SAN is always difficult, but the return
on investment is better than other technologies. Once SAN is deployed, scaling
it is not difficult and the advantages it brings makes it inexpensive,"
Subramaniam said.
In this maze of technology, IP-SANs are gathering momentum and IDC predicts
that 23 percent of the overall storage market would be on IP-SAN.
Using fiber channel, iSCSI, or IP in a SAN depends on the policy of the
company. Storage experts say fiber channel being a mature and robust technology
should be used for mission-critical applications. However, iSCSI gives the
flexibility of using the Internet for access, and hence comes out to be cheaper.
Data Center or Own Network
Whether to outsource data storage to a third-party data center or to have a
network within the campus is a business decision an organization has to take.
"If a company is looking for short-term data storage and the life cycle is
short, a data center serves the purpose. But if long-term storage is being
planned, a serious thought must be given to own the network," said Leong.
When the frequency of data access is higher and everything needs to be
online, own storage network is desirable. But the trend today is of a hybrid
model. "Most companies have their data disaster recovery network with a
third party and for the current flowing data they have their own
environment," added Chugh.
The rise in data volume has transformed storage from being a tactical
decision to a strategic one. Companies now see storage network as an asset and
the investment made in it as more of capex than an additional revenue burden.
Security and Disaster Recovery Storage
When storage network is within the organization, it is less prone to
security threats. But as soon as data leaves for a data center over a public
network, there is a need to have some checks. "There has been no report of
a fiber channel network being hacked, however, security is on the top of the
agenda when a CIO is planning his storage network," said Subramaniam.
In the Western and many South Asian countries, there are regulations and
mandates to put in place security mechanisms in data centers and disaster
recovery sites. Though there are no such legal bindings in India, the storage
vendors have also woken up and are putting multi-layered security protocols in
their products.
Improved Security with Zoning
Zoning allows automatically or dynamically arranging fabric-connected
devices into logical groups (zones) across the physical configuration of the
fabric. These zones can include selected storage, servers, and workstations
within a fabric. Information access is restricted to only the 'member' devices
in the defined zone.
Although zone members can access only other members in their zones,
individual devices can be members of more than one zone. This approach enables
the secure sharing of storage resources-a primary benefit of storage networks.
In addition to improving security, zoning can also help to simplify management
of heterogeneous fabrics, maximize storage resources, and segregate storage
traffic.
A secure fabric operating system is a complementary feature to zoning. Secure
fabric operating systems run on SAN infrastructures, and offer policy-based
security. These policies allow you to customize security uniquely to your needs.
Secure fabric operating systems help block unauthorized fabric-wide management
changes and fabric setting changes, help control server-to-fabric connections,
prevent users from arbitrarily adding switches to a fabric, and protect
communications between switches and management consoles.
There is no technology constraint in having security software in storage
devices. But there is a need for a conscious policy to cover everyone as far as
security of data-be it within the campus or at a data center disaster recovery
site-is concerned.