We are about to set up an automated not-for-profit TV channel and need an inexpensive way to store all our MPEG video files that will be played. We basically need space in the order of 1TB to 2TB. So, how can we do this inexpensively without going to the big (and expensive) guys? Can we build it ourselves using off-the-shelves components? It would be nice if we can mirror the data for protection and we're smart enough to setup our own network if needed.
There are a number of ways to do this that can meet your goals.
1) As long as you have the talent to provide your own sweat and support, you can get big storage on the cheap. Assuming you have Linux experience (don't fool yourself if you don't), build a Linux file server and put two terrific open source software products on it: the XFS file system from SGI, and Sistina's Linux logical volume manager (LVM). XFS gives you an optimized file system for big files and the LVM gives you the disk management, redundancy and scalability you are looking for.
Make sure you use "beefy" hardware to get the I/O throughput you need. Buy a Fibre Channel HBA with Linux support. Then buy an off-the shelf FC JBOD disk cabinet with the capacity you need. You can connect the HBA to the JBOD point-to-point or with a hub. A switch is overkill for your needs. Think about using RAID 0/1 for this application. It takes twice as much raw disk capacity, but the performance and data protection fit your requirements.
2) Buy an off the shelf "big" NAS system from a "small" NAS system manufacturer. There are a number of vendors trying to "move up" the food chain and you can get good sized systems for bulk storage for low cost.
3) Buy an inexpensive off the shelf RAID cabinet such as Adaptec's Durastor and connect it to a server. There are many vendors that a Google search will yield.
4) Look for multi-channel ATA controllers, such as the Escalade product from 3Ware. Make sure you plan your cabinet and cabling needs carefully.
5) Write to vendors and tell them you want to be a beta customer for new products. What you lack in "market visibility" you make up for by actually using and testing the product. But don't do it if you aren't serious about really working with the product. This requires a certain amount of luck and timing and don't feel put down if the company can't work with you for one reason or another. You need to understand what the cost will be at the end of the beta period because beta does not equal free but your price should be discounted. You also need to be prepared to be used as a reference by the vendor.
Editor's note: Do you agree with this expert's response? If you have more to share, post it in one of our .bphAaR2qhqA^0@/searchstorage>discussion forums.
This was first published in July 2002