I recently worked with a small business that was looking for a way to organize the data storage on the network. They had a number of servers, including two Windows 2000 domain controllers, a terminal server, Web server, and two dedicated file servers, as well as about 10 workstations. There was no lack of free disk space on the network; the problem was that it was spread out all over the network.
I call it the "it just grew that way" syndrome, and I see it on many small to medium networks when companies start out with just a few computers and add more as they add employees and/or take on more projects. The more the company grows, the more scattered the data gets.
It's much easier to plan a good, scalable data storage strategy from the beginning. But does this necessarily mean putting all your eggs (or in this case, all your data) in one basket, or does distributed storage make more sense? Certainly, having the company's data in one place makes it easier to manage, secure and back up. But you'll want a good fault tolerance plan to protect against data loss.
There is no "one size fits all" solution — even for all businesses of the same size. Most companies, once they outgrow the "store it wherever it happens to fall" stage, move to file servers with some sort of RAID array. This is the time to plan for anticipated growth.
When you think of scalability in regards to your RAID solution, you may think only of storage capacity: can you expand the disk capacity to meet your future needs? In fact, this is only one aspect of scalability, but it's an important one. How many disks can be added to the array? You can connect up to 15 disks per SCSI channel (SCSI-2). That might sound like far more than you'll ever need, but remember that 10 years ago, a one-gigabyte hard disk seemed huge. We didn't anticipate the huge video and media files that are commonplace today, and as applications become more sophisticated, we can't predict the sizes of the files they will create.
For greater capacity, consider going with fibre channel instead of or in addition to SCSI. A fibre channel loop can support 126 devices. Another advantage of fibre channel is that it supports longer cable distances. As your business grows and you add more devices, you may also need to spread them out more. Whereas SCSI works best with cables no longer than 3 meters, fibre channel allows for more than 30 meters between devices, and a fibre channel Storage Area Network (SAN), using dark fibre, can support links of several kilometers in length.
In selecting a solution, you should consider whether the hardware vendor has planned ahead to support new, faster, higher capacity disks in the future. Will new hardware or software be required to use the newer disks?
The nature of your business may dictate that you need scalability without downtime, so you'll want to find out whether disks are "hot addable": if you need more capacity, can you add disks to the array without taking it offline? Not so long ago, adding a disk to an array required backing up all your data, adding the disk, reformatting the array and then restoring the data. Most modern RAID solutions provide you with configuration tools for changing or adding disks dynamically. Again, fibre channel offers advantages with its better support of hot plugging.
Check out not only whether you can add disks dynamically, but the configuration in which disks must be added. A SCSI array in which the data is placed on one disk with parity information spread across five disks will usually require that you add six disks every time you want to increase capacity. With fibre channel, arrays can be configured so that you only need to add one disk at a time.
Fibre channel technology and SANs
Fibre channel technology is the basis for SANs, the next step up in scalable storage solutions. A SAN is a group of networked storage devices that can be directly accessed from all of the servers on the network. As your business grows, speedy access to data becomes more important, and a SAN provides a fast, switched connection.
Best of all, SANs can grow with your business and your network. Additional storage devices can be added to the SAN as the need arises, and with little or no downtime. Additional servers can be added to support more users (without adding more storage capacity unless/until it's needed). Switches can be added to prevent deterioration of performance as more users and storage devices are added.
A SAN can start out small — just one SAN-attached storage device and a couple of servers. The SAN configuration can grow as your storage needs increase. You can attach backup tape devices to the SAN, for faster backup.
What about Network Attached Storage (NAS)? While a SAN in essence moves storage to its own network and eliminates the file server, NAS builds on the file server concept but simplifies it by removing all the unnecessary functions of the operating system. A typical file server runs Windows, NetWare or a UNIX/Linux operating system. This creates extra resource overhead and makes the file server vulnerable to security breaches and attacks that target that OS. A NAS device uses a stripped-down OS that makes it more secure and faster than a regular file server, although it usually uses an Ethernet connection to connect to the network like the typical file server.
NAS devices are often "turn key" appliances that are easy to plug in, set up and operate. They are usually less expensive to implement than SANs. However, they have traditionally suffered from lower performance and more limited scalability. Since they are self-contained appliances, it's been difficult to add capacity. However, new technologies, such as switching devices that let you distribute data across multiple NAS devices, are changing this and the convergence of NAS and SAN is making the former more scalable than ever.
Choosing the right storage solution for your growing business isn't a "no brainer." You have more choices than ever, and scalability and performance must be weighed against ease of use and the ever-present budget considerations. But spending a little more now to plan for future growth can actually save you money in the long run, so take the time to ask the right questions about scalability when you compare storage options.
Debra Littlejohn Shinder, MCSE, MVP is a technology consultant, trainer, and writer who has authored a number of books on computer operating systems, networking, and security. Deb is a tech editor, developmental editor, and contributor to over 20 additional books on subjects such as the Windows 2000 and Windows 2003 MCSE exams, CompTIA Security+ exam, and TruSecure's ICSA certification.