Project History
The Media Server Project has actually been in the works now for well over three years. Limitations on time, energy, technology and most importantly cash, resulted in the project being pushed to a back burner since my initial inspiration back in 2004-2005 time frame. Back then I was spending an extensive amount of time building a web-based interface for all of our audio CDs, the bulk of which I had ripped in Microsoft's WMA format and stored for easy access on a file server. But that's a whole different project to be discussed at a later date. This article is focused on applying the same basic principles, but to DVDs instead of CDs.
Project Requirements
For most projects like these, I prefer to do a lot of research beforehand to make sure my money is well spent and that I'm going to get the end result I seek, or at least as close as I can to it. This project was no different in that regard. There were three very basic, but essential, issues that I needed to resolve in order to make this project a success:
- Media Ripping Requirement: In addition to decrypting the DVD files, I needed to find software that would maintain the original audio and video quality, which meant absolutely no compression.
- Media Storage Requirement: Given my "no
compression allowed" requirement, each DVD was going to eat up
approximately 6-8 gigabytes worth of space on a file server. Even for a
moderately large DVD collection, the storage space required for raw
format DVD files quickly becomes overwhelming.
- Media Playback Requirement: Since I planned to leave the DVD files in their original .VOB format, I required a user-friendly application that would recognize this format for playback.
Thanks to a few media server pioneering articles online, I quickly discovered that the previously-noted technical limitations would no longer pose a problem. Because of the technological advancements in both software and hardware over the last few years I now have the following options available:
- Media Ripping Solution: DVD Fab's HD Decrypter software allows for full decryption with no compression. Installation and configuration is very straightforward. Plus, compared to previous applications I tried that required 1-2 hours to rip a single movie, DVD Fab averages around 20-30 minutes to rip an entire DVD, including menus and other bonus features.
- Media Storage Solution: The prospect of paying for enough space to house 1000+ raw format DVD files on a file server somewhere used to make my eyes bulge. It simply wasn't feasible. This is no longer the case however, since SATA
hard drives are currently available in up to 1.5TB capacities at a
price point of less than ten cents per gigabyte.
- Media Playback Solution: Finding
software that recognized .VOB format back in 2004 was virtually
impossible. Thankfully, this is no longer the case. Microsoft's own Media Center, a
built-in feature of any Windows Vista Ultimate Edition, now has full support
for DVD playback. As an added bonus, the kind folks over at My Movies
have also created a custom plug-in that works with Media Center and
offers an even richer experience for avid DVD collectors.
Project Hardware: The Chassis
Since the bulk of the existing server systems I own are Supermicro, it's probably no surprise to anyone that I headed back to them for this project as well. I have used their hardware almost exclusively in the past with fantastic success. Their systems not only perform well, but they seemingly last forever and are extremely stable in just about any application. Due to the massive amount of storage required in this particular case, I needed a chassis that would support a large number of drives, not only to house the DVD files, but to allow for future expansion as well.
After reviewing the available options online along with a few phone calls to Supermicro directly, there was no question that the SC846TQ-R900 was the perfect solution for my needs. The 4U chassis compactly manages storage space for up to 24 hot-swappable 3.5" SATA-compatible drives, as well as two additional slots for a separate system drive and DVD-ROM. Air flow has been beautifully engineered into the design of this case offering five high-capacity, RPM-adjustable fans as well as a custom air shroud, all designed for direct cooling where it's needed the most.
Project Hardware: The Motherboard
Motherboard selection was another process entirely. I wasn't necessarily bent on Intel vs. AMD. But for the sake of familiarity, I opted for an Intel-based board again. Supermicro has a lot of motherboards available so I was doing everything possible to sift through all the options. Based on what I have read, the biggest potential bottlenecks for a streaming media server are memory and bandwidth. A multi-processor configuration is nice to have but is not a necessity from a performance standpoint. However since I typically overtask my files servers by forcing them to wear multiple hats at once, I opted for a dual-CPU motherboard. After much deliberation and comparison research I opted for the X7DWE motherboard.
Because this system will be racked, and is certainly not a gaming platform, I wanted a motherboard with onboard video in order to save the expansion slots for additional RAID controllers or other requirements. I also preferred to get as many PCI-e slots as I could since most of the newer cards are going to take advantage of that format and the inherent bandwidth that comes with it. Total memory requirements were not necessarily written in stone so I was somewhat flexible there. This particular motherboard has four slots that support 8GB DIMMs, giving me the option to expand up to 32GB of memory if needed.
Project Hardware: The Hard Drives
Storage hardware was probably the easiest decision I had to make. Ideally, I needed the biggest drive available to take full advantage of chassis' 24 slots. Although experiences may vary, my particular experience with
Seagate hard drives has been generally positive. And since Seagate was currently the only manufacturer to offer SATA II hard drives in a 1.5 terabyte capacity, my decision was easily made. Again, my focus here was on total capacity not speed. So even though these drives didn't rate quite as high from a performance standpoint as their competitors, they did offer 50 percent more capacity, making them the obvious choice. I plan to purchase eight of these drives up front and expand with more drives at a later date if and when space is running low.
In addition to using Seagate's 1.5TB drives for storage, I decided to take advantage of the internal 3.5" bay for a single system drive. The trade off to a single drive scenario is that there is zero redundancy. But housing the system drive internally also means not using up critical space in the external drive bays, space that will be much better served for storage. Although it wasn't entirely necessary, I decided to give the main system drive a slight boost in performance by purchasing one of Western Digital's 150GB, 10,000 RPM VelociRaptor drives. I've had good luck with the Raptor drives in the past so this also seemed like a no-brainer to me.
Project Hardware: The RAID Controller
I went back and forth for awhile between various RAID controller manufacturers, exploring products from Adaptec, 3Ware and Areca in the process. From my experience and research it seems that all three of these companies offer well-designed solutions for aftermarket RAID. Unfortunately, none of them offered the same capabilities as my current RAID controllers, manufactured by Ciprico using the RAIDCore chip. In a true example of terrible timing, my attempts to purchase more RAIDCore controllers were foiled when I discovered that Cirprico had filed for bankruptcy.
Despite my enthusiasm for the RAIDCore products, all attempts to contact anyone at Ciprico regarding future hardware production or support resulted in no response. Bankruptcy was obviously a pretty serious blow to Cirprico, as a company, but even worse for the RAIDCore technology itself. As an avid fan of RAICore since it's inception back in early 2000, I have followed and purchased various versions of the product as it changed hands, each time seeing marked improvement in features and performance. As far as I'm concerned, no other RAID Controller comes close to when you compare features to price.
Regretfully, I returned to 3Ware and Areca, dismayed at the prospect of using what I believed to be a lesser product. As a last resort, I decided to dig around online to see if any web-based hardware distributors still carried the RAIDCore 5000 series controllers in stock. Thankfully, I found a company that had quite a few of the RC5252 8-port controllers still available. Given that the motherboard supports up to four separate PCI-e devices, I saw no issue with ordering three of these controllers, giving me full future support for all 24 drives.
Before I get any comments based on the utter stupidity of purchasing a product that offers zero support, you should all know that I've been using these cards for awhile now and with great success. Based on that, and the fact that reports have been released indicating that Ciprico has officially sold the RAIDCore rights to Dot Hill, I suspect I'll be seeing future versions of the technology soon enough, along with extensive support.
Summary
I apologize up front because I originally began writing this article back in October so I'm a little behind in when it comes to updating. In the two months that have passed since this article's creation, I managed to successfully purchase all of the above hardware and the media server now sits fully racked and is performing even better than expected.
For those who may be interested, I plan to follow this article with a series of others, outlining hardware, software and other configuration details concerning this project. Among other items left to discuss is the front-end Media PC used to interface with the media server itself. Check back soon for further details on Project Media Server!
Related Articles
RAIDCore: A Brief History
DVD Ripping Utilities