If there is one thing that remains a constant in e-business, it's that e-business itself creates mountains of data in the forms of customer data, transaction data, sales data, product data and operational data. Analysts are pegging the annual growth rate of corporate data at anywhere between 50 to 150 percent. With the increasing adoption of data warehouses, document and content management systems, ERP, supply chain management and product data management, even the upper end of that range may prove to be too conservative. In the final analysis, whatever your current storage capacity needs are, it's a good bet that next year -- you'll need more.
It wasn't too long ago that adding more storage capacity was a straightforward proposition: buy more disks for the server. But as more and more of the enterprise goes online, with larger file sizes and rich media data types such as video, the need for storage grows almost exponentially, driving demand for solutions that are scalable, available, flexible, non-disruptive, secure, integrated into existing networks, and above all, are easy to manage. Oh, and did we mention affordable?
As a result, we are seeing a tremendous growth in Storage Area Networks, or SANs. SANs are dedicated high-speed networks that relieve the main network from the tasks of data storage, management and retrieval. The Enterprise Storage Group recently predicted a 1000 percent growth rate for SANs. Gartner (May 2001) sees spending on SANs topping US$35 billion by 2003, and IDC predicts that by 2005, 70 percent of corporate data will reside on SANs.
The evolution of SANs
To understand the value of a SAN, it's useful to review the evolution of storage technology. Over the last decade, as networked NT and Unix servers proliferated, data was maintained on hard disks within each network server. The problem with this -in-server storage" was that it consumed server resources every time data was accessed, saved, backed up or restored.
With the advent of RAID (Redundant Array of Independent Disks) technologies, which was designed to increase throughput and/or reliability, storage devices began to move out of the server and become independent entitites -- attached to the server but with their own file systems and management software. This model, known as direct attached storage or DAS, took the burden of processing storage and backup requests off the network server, and was an evolutionary step in the right direction. However, it was expensive and required devices to remain tethered to the server.
NAS (Network Attached Storage), currently in wide use, addresses the shortcomings in DAS by cutting the link between the storage device and the server altogether and making it part of the network. NAS devices are essentially dedicated storage servers that can be accessed from anywhere on the network, can be shared by multiple clients and servers, and provide high capacity (up to 30 terabytes) at a relatively low cost. Moreover, it's relatively easy to add capacity simply by adding more NAS devices.
NAS devices have moved from departmental to enterprise-wide use because they are inexpensive, easy to install and configure, and can be expanded to accommodate growing data needs. NAS devices connect to the Local Area Network (LAN) via Ethernet and use standard protocols such as TCP/IP or IPX. Because each device has its own file manager (known as a -filer"), it enables data to be available on the network even if one or more of the main servers is offline.
However, the NAS approach does have limitations, particularly at the enterprise level. NAS devices deal with data files, such as emails and Web pages, but are not designed to handle discreet chunks of data, as is the case for e-commerce transactions or streaming video. And although installing a NAS appliance relieves the main application servers of storage management tasks, the network itself is still burdened with the overhead.
This is where SANs come in. A SAN is a separate network of storage and backup devices designed to handle high levels of data traffic and offload them from the main network. They use a high-speed connection, most commonly Fibre Channel (FC), to read and write raw data to and from the various storage devices at data rates approaching 1 Gigabit per second, some ten times faster than a typical Ethernet LAN. Devices are hot-swappable on the network, thanks to Fibre Channel switches which make up the -fabric" of the SAN.
Choosing between NAS and SAN is hardly an either/or proposition. In fact, the two technologies are already converging, and that convergence will accelerate over the next few years, resulting in solutions that combine the file-orientation and ease of use offered by NAS with the speed, availability and scalability of SANs.
When is a SAN in the cards?
SANs are not for everyone. The are expensive, typically priced from $250,000--although a recent multi-vendor consortium known as the Affordable SAN Initiative promises to bring the costs down to under $50,000 for an entry-level SAN. Whether your business needs a SAN depends on many factors:
- Company Size: Larger businesses that need to scale storage quickly and/or maintain data warehouses are natural candidates for a SAN. However medium-sized enterprises that are growing repositories of customer and transaction data may also benefit.
- Nature of applications: Enterprise applications such as CRM, ERP, SCM, OLAP, and OLTP that deal in discrete data blocks rather than files are all good candidates for SANs.
- Type of data: The need to store large amounts of video and image data mitigates in favour of a SAN.
- Industry: Data and transaction-intensive industries such as financial, retail, entertainment, and health care will benefit from SANs.
- Throughput requirements: Organisations that need to move data quickly from one point to another will benefit from the ability to offload that data onto the SAN.
Kathy Hahn and Andy Keith are lead consultants at Juldee, a strategic marketing services firm specialising in xSP, Mobile, Storage, Internet Security and Enterprise Infrastructure.
