New, demanding computer applications and business process re-engineering
have contributed to fundamentally change the nature of data in
distributed environments, causing a shift in how users store and
manage information. On the one hand, the new requirements tend
to focus on quick and reliable access to large amounts of information,
as opposed to data processing speed, modifying the traditional
notions of performance and data accessibility. On the other hand,
the data explosion and the need for a pervasive access to mission-critical
data scattered among heterogeneous platforms, have increased the
complexity and cost of managing storage in corporate environments,
calling for increased connectivity, scaleability and manageability.
These new objectives must be reached in a context of reduced manpower
and tighter budgets.
A new breed of product, the Network-Attached
Storage Server (NASS), is about to address this dilemma, offering
the best solution for these sometimes conflicting objectives.
A network-attached storage server constitutes a large storage
pool shared by many clients, available to all as a network resource.
It is the culmination of two concepts: the direct attachment of
storage to the network, and the consolidation of storage resources
in a powerful Intelligent Storage System. It lays the foundation
for what is a fundamental departure from the present computer-centric
world, to a storage-centric architecture that puts data where
it belongs, at the center of the computing environment. Through
very high data accessibility, autonomy and direct connectivity,
NASS helps break an important barrier, placing storage (or data)
at the center of the computing environment (Figure 1).
Figure 1
The User's View
In a series of surveys conducted by Peripheral Concepts, Inc.,
administrators and managers of information systems were asked
to rank the characteristics they considered most important in
the acquisition and management of their storage. Five features
- Accessibility, Scaleability, Performance, Connectivity, and
Manageability - together with price and cost of ownership, were
consistently ranked at the top of the list of most wanted features.
Figure 2 shows one typical survey result.
Figure 2
THE USER's VIEW
Importance of Storage and Storage Management Characteristics
Network-Attached Storage Servers present
a new approach to responding to these requirements. In many user
organizations, there is a growing dissatisfaction with traditional
ways of providing shared storage over the network. A study conducted
by Strategic Research Corp. - "Driving Server Centralization"
- found that the traditional method of using a file server as
a storage repository has become the least favored solution. The
majority of users, according to the study, now think about shared
storage on the network as a task that should be performed by a
specialized, dedicated storage server or a specialized network
device. In order to improve performance, achieve greater uptime,
and reduce the cost of storage management, 46% of the sites surveyed
have already installed a dedicated storage server.
What is a Network-Attached Storage Server?
A Network-Attached Storage Server (NASS) is a self-contained,
highly-accessible, intelligent storage system that is independent
from the server's operating system. It uses file system access
protocol and standard network protocols to communicate directly
with clients on the network. This makes it a fully integrated
and dedicated storage solution that connects to the network, providing
a network file storage repository as a shared resource. To applications
running on the network, the Network-Attached Storage Server looks
like an ordinary server. To any client, it simply looks like a
large hard disk.
NASS provides autonomy from a central server
(or central processing) unit, and optimizes the cost/performance
ratio for a wide range of products covering multiple applications.
It consolidates the storage of many application servers into a
single storage service offered by one manageable pool of physical
storage. Through consolidation of larger capacities, options that
would otherwise be prohibitively expensive become practical, permitting
high levels of performance, scaleability, and manageability. NASS
also raises the standard of accessibility and connectivity to
levels never reached before in the open system environment.
By overcoming the limitations found in traditional
distributed storage systems, NASS responds to the new storage
and storage management requirements, as summarized in table 1
and discussed below.
Table 1
MARKET REQUIREMENTS AND NASS CONTRIBUTION
Performance
New applications have created new requirements for high speed
transfer of very large files. General purpose servers and most
operating systems were developed for fast processing in a multi-user,
multi-task environment, and are a poor match for handling such
requirements. Performance problems and traffic bottlenecks develop
when data is transmitted from one server to another, or more accurately
from one storage device attached to a server to another storage
device at the other side of the network, which is also connected
to a server.
With a dedicated, network-attached controller, NASS provides greatly
improved performance. The ability for the controller to connect
anywhere on the network, allows network performance to be balanced
by placing storage close to the users who need to have data readily
accessible. This architecture is particularly effective when the
network is segmented by bridges, routers or switches (Figure 3).
Figure 3
Other potential performance improvements are related to the optimized software component of NASS. They are derived from optimized cache algorithms, the use of separate control and data paths allowing for high throughput rates, or the design of admission control mechanisms guaranteeing bandwidth for high priority data streams and time-dependent events.
Accessibility
Mission-critical data, now deployed in most organizations on distributed systems, must be accessible at all times to maintain user productivity. Accessibility is a continuum that ranges from component resiliency to a full system and connection resiliency, including automatic fail-over to an alternate server or path.
Currently, availability requirements in many organizations are for what is now referred to as 24 x 365 operation, an availability level that must also protect against catastrophic failures, where data and working environments are replicated in a remote location to recover from site disasters.
In addition, a NASS may provide full data and application accessibility. For data, beyond redundancy, accessibility implies automatic fault detection, isolation and recovery, on-line reparability, and complete system restoration after a failure has been fixed. For applications, it means a thorough testing program that includes fast recovery from unexpected conditions occasioned by system component failures.
Connectivity
In today's heterogeneous computing environment, the simultaneous connectivity of the shared storage system to several platforms or operating systems and its ability to serve a variety of file systems and connection types has become an important requirement. Connecting and disconnecting needs to be easily performed without disrupting the network nor the user.
There are two choices today for expanding storage: add storage to an existing server, or add a new server. The first alternative requires bringing the server down, installing a host adapter, attaching the device, installing a driver to communicate with the host adapter, and running tests to ensure that there are no incompatibilities. Adding a new server involves costs associated with reconfiguring the system, connecting to the Small Computer System Interface (SCSI) ports, and providing or attaching drivers.
NASS offers a third choice that plugs in in minutes, and is not affected by upgrades of servers, operating systems or applications. It appears to the application as another server. It can be attached without shutting down the network, and requires no changes to existing file servers. For the administrator, it simplifies the task of configuring, testing, and ensuring that independently selected modules integrate harmoniously. NASS can be located close to the clients and then easily moved and re-installed on the network as configurations change.
NASS ties to a network protocol and topology. This makes it available to the entire network, supporting heterogeneous platforms and operating systems. As an example, NASS can serve most of the UNIX market if it supports the Network File System (NFS) file access protocol. With NASS, one can integrate additional storage at any location without changing the overall storage strategy; the exact amount of storage needed is plugged-in for a modular incremental growth. The system platform independence and direct network connectivity allow corporations to retain their prior investments in their storage devices when they change environments, optimizing their utilization.
Scaleability
The surge in the number of users and in the amount of data collected and analyzed, has created a tremendous growth in the demand for additional storage capacity. Data explosion is further fueled by new applications involving increased integration of images, voice and video. In an open system environment, disks are added or reconfigured once a month, on the average.
Many available products limit their scaleability to enclosures presenting mounting space for several controllers, disks, tapes or optical devices. While these systems allow the user to start with a minimal configuration, and grow by adding devices to meet requirements, they cannot easily integrate other growing needs.
Scaleability, as a criterion for evaluating a NASS, must allow for growth in capacity, performance, connectivity and accessibility, without affecting on-going operations. Scaleability includes adaptability and portability, as well as the ability to convert to a full range of products covering multiple applications. This implies that the implementation finds a common hardware and software denominator capable of growing with expansion needs.
Manageability
In a distributed environment, the real cost of storage is not in the hardware, but primarily in the labor involved in managing storage, and in the productivity lost due to network outages or slow data access. The management of storage represents one of the most significant costs incurred in the administration of networks. Studies conducted by Peripheral Concepts show that while every megabyte of commodity disk added to a UNIX server costs less than 60 cents to purchase, the company incurs an additional 3.5 dollars per megabyte each year in labor and lost productivity to manage this storage. Increased manageability has become a priority in all distributed network environments. Storage manageability requirements include automated procedures to relieve the administrator from the burden of individually monitoring each node, as well as the integration of storage management chores such as backup and restoration of data.
NASS manageability starts with the ability to easily install and use large amounts of storage. Storage must also be groomed, upgraded, modified, and reconfigured for optimum performance. Better storage systems can accomplish this without disrupting the user, that is, without forcing any interruption in the access to data. Storage management involves the protection of data (backup, archiving, remote vaulting) and the management of media, in compliance with policies and procedures. Equally important are the efficient and easy restoration of data in case of errors, failures or disaster recovery. The extent to which a NASS can integrate and automate some or all of these steps, determine its position in the hierarchy of product classes described below.
Storage has long been regarded as a necessary evil, and a major cause of system downtime and network bottlenecks. Network-attached storage servers open a new era, one where storage becomes an autonomous reliable entity at the center of the computing environment. They constitute a major step forward towards an ultimate goal, one that allows users around the world to share files, images and video clips, without concerning themselves with data getting lost or corrupted.
Obviously, not all network-attached storage servers can, nor need to, assume such a central role. It is important to differentiate between several classes of products to determine which best fulfills the specific user's needs. As always, products with similar names may offer different functionalities, with corresponding price variations.
Generally speaking, one can differentiate between three major classes, with the understanding that no hard boundaries exist between the product classes, and that the definitions evolve with time, as pressure increases on vendors, and features are made affordable by advances in technology.
A class-1 NASS is resistant to the most common failures, as disk drive, power supply, and cooling system, but it is still vulnerable to less common component failures, and most often must be turned off for planned maintenance. Its connectivity is limited to one network protocol (Ethernet, for example) and one file system.
A class-2 NASS offers a resiliency that leaves no single point of failure, and allows all field replaceable units to be swapped while the system continues to operate. It incorporates some performance features, such as faster write for RAID configurations, complex caching algorithms, and/or dedicated data paths. The system can be scaled many ways, allowing easy growth in performance, resiliency and connectivity. An integrated backup solution is usually offered as an option, and administration of all network-connected storage can be performed from one central location.
Features contained in a class-3 product include connectivity to several network topologies and file systems, extending to proprietary systems and mainframes. Performance is improved through automatic load and path balancing, and high speed transfers between storage domains assigned to different file or application servers. Accessibility is enhanced by an important feature that assures disaster protection, where an entire site may take over upon the failure of a complete system. Failing components are self-diagnosed, reported to a remote service site, and eventually corrected or replaced before they become inoperative. Storage capacity can be scaled to over one terabyte within an enclosure. The same hardware and software elements constitute the basic core system capable of covering a wide range of applications and different data types.
Table 2 summarizes the three major classes of network-attached storage servers.
The merits of network-attached storage servers
have attracted interest from the major system and storage vendors,
in a wide range of implementations. StreamLogic's StreamLink is
an example of a Class-1 product, Symbios' MetaStor SH1000/4000
an example of a Class-2 product, and EMC's Symmetrix Network File
Storage (SNFS) an example of a Class-3 product.
Storage acquisition and management is undergoing great changes, stimulated by new applications, a growing need for a high level of data accessibility, and the constraint and management complexity of heterogeneous platforms and operating systems. Network-Attached Storage Servers (NASS) represent a logical evolution in the storage paradigm. They combine the unique advantages of direct network attachment and storage consolidation.
Direct network attachment allows NASS transparent connection to any client on the network. It enables workgroup-centered, shared data to be placed next to its users, and enterprise-wide data to be accessible directly through a high speed network backbone. Creating a centralized storage pool allows larger purchases, permits flexible deployment of capacity across servers, and provides the ability to share expensive technologies. By integrating storage management within the centralized storage, the burden of management is moved from the client or the application to the system.
Most storage solutions offered today are too fragmented and slow the progression to newer applications. Storage needs to be put at the center of the computing environment to allow efficient access to data. Higher classes of Network-Attached Storage Servers elevate storage from its present status of a peripheral to a major participant in network traffic, facilitating the speedy transfer of data in a variety of formats.
By: Farid Neema
This Report was produced by:
PERIPHERAL CONCEPTS, INC.
351 Hitchcock Way, Suite #B-200
Santa Barbara, California, 93105
Tel: (805) 563-9491
fneema@periconcepts.com