Recently, I received an email message from our IS department stating that our editorial server's disk was full and requesting that we delete "unnecessary" files. As a former IS manager, I realize the importance of managing disk space, but IS had installed this editorial server only a few weeks earlier. Now that I'm an end user, I find these messages annoying. Heck, all my files are necessary.
Ever-Increasing Storage Needs
Surely, someone in IS took our disk requirements into consideration before installing our new server. Nevertheless, our storage was full, and some poor IS folks were backing up the existing storage, adding disk drives, restoring, configuring, and testing. While the editorial staff was home eating dinner, the IS folks had to tell their families they'd be home late—again.
Our company has dozens of NT-based file servers. Each server has a set of disk drives to manage. At any given time, some disk subsystems are nearly full and others are nearly empty. Our IS group would love to be able to instantly take storage from an underutilized server and give it to an overextended server. Unfortunately, the disks are in separate servers, making this task difficult.
SANs to the Rescue
There is another way. Storage Area Networks (SANs) let you centralize your storage.
SANs are like LANs, but for storage. Imagine removing the disk subsystems from 12 servers and consolidating them into one cabinet to create a SAN. Because the SAN would house the combined data for all 12 servers, each server would need to have only enough storage to boot NT. Each server would have a logical piece of the SAN to handle all the data needs of its attached users.
SANs benefit NT administrators in several ways. First, SANs let you manage backup and recovery of all your servers from one location. Today, to make your backup window, you attach tape drives to multiple servers. Usually, that practice means you have multiple tape drives and multiple backup schedules for multiple servers. With a SAN, you can manage all your storage centrally. You can use one tape library to back up your entire organization's data. If you use replication, you can make realtime copies of the data, allowing the backup system to run 24 X 7.
Second, a SAN allows more flexibility in assigning storage to departments. With a few clicks of a mouse, our friendly IS folks could simply assign more storage to our editorial server. No new disk, no backup, no testing, and no excuses for missing dinner.
Third, a SAN lets you add and replace storage easily. To add storage, you simply put another logical RAID module into the SAN cabinet. If a disk drive goes bad, the SAN's built-in alert system notifies you. You simply remove the offending drive and add a new one. The SAN will rebuild the drive automatically.
As a result, you don't need to disrupt users, because all drives are hot-swappable. Because a SAN assembles drives into logical modules, a SAN will let you mix and match different drive densities (e.g., 9GB, 18GB, 36GB) in the same cabinet.
Fourth, a SAN frees up network bandwidth for applications that require massive file movement. For example, suppose you have a terabyte of data in an operational database and need to update a data warehouse daily. In a typical configuration, you might move hundreds of gigabytes of data across your network. You might have to reengineer your network to accommodate such an application. With a SAN, you transfer data only within the SAN, and this transfer doesn't interfere with the rest of the network.
Finally, a SAN lets you separate your network's compute engines from the storage engines. Because you store no business data within a server, you can upgrade servers without affecting the SAN. The server upgrade process is simpler and less risky because it affects no business-critical data.
Home for Dinner
This technology is a promising way to solve several storage problems and to make end users happy. Next time you're missing a dinner to watch disk-drive lights blink, consider a SAN for your organization. Your users will thank you.