Back to it this week. I had two project drives become corrupt this week. I was able to save the data but it was a pretty painful process. In real need of something bigger and more robust.
Power supply arrived yesterday, installed. Still looking at CPUs.
It's still winter, and I am making progress :-) Getting the first 10 TB of 7.2k RPM drives was a huge win. CPU and power supply next week!
5 of the 2 TB drives arrived today. I'll start putting everything together this weekend. I'll supply pictures as I go.
Finally! 10 2 TB sata drives coming my way. I will resume work on this project ASAP. So stoked.
I had to pick up one of those fold out tables so that I could set up a workspace for this project in my basement. Part of what was slowing me down was just having a dedicated space to do the work. I'll be ordering the power supply and mother board/memory this week. I have enough 3.5 inch drives kicking around to build something that I can at least do a proof of concept with.
I'm lining up a source for the drives that I'll need to fill the ten bays that I have. I'm looking at picking up dispositioned drives from Equallogic. Apparently, they test thousands of drives from different vendors before qualifying them for use in their storage arrays. If the drive fails and is out of spec, it doesn't mean its a bad drive, it's just not good enough to go in one of their high priced arrays. It's probably more than adequate for my needs, and certainly cheaper than retail.
With my latest reality TV sizzle turned into my production partners, the time has come to turn my attention back to my editing infrastructure and get this thing built so I can have a home for the archival of some of my older projects.
The unfortunate thing is this SAN project is now competing for time with my documentary project, yet, since I'll be shooting a fair bit of RAW footage for the doc, I'm going to need the space and can use the SAN. No, I can't afford to buy a commercial one that will give me the performance and size that the one I'm building will have. I also need the money for some new camera gear, college tuitions, …. life.. This should be quite the interesting balancing act...
Too much time went by between deciding on and actually ordering the server chassis I mentioned in my previous post, and they were out of stock by the time I was ready to buy.
I instead settled on:
which had about the same specs but was a little cheaper.
It arrived last week and despite the plastic drive trays, a fairly common feature of chassis in this price range, the build quality is not that bad.
I've been slowed a bit on my build progress. Distracted by other projects at the moment. But I have read and watched a variety of articles and videos on this. I've settled on the following server chassis:
The most likeable things about it are:
Made a few changes to my current set up that seems to have fixed some of the audio latency issues I was seeing.
I had an old M-Audio firewire solo that was at the end of the drive chain as its FW 400. This is an input device for microphones and instruments, but I was also using it to drive my studio monitors. I was getting this audio latency issue that was making it look like things were out of sync and I wasn't entirely sure where it was coming from, but there were other things that were feeling slow, so I assumed that all the drive chaining had something to do with it.
After installing the latest Mountain Lion update, the FW Solo stopped working reliably and there were no more updates from M-Audio on the driver as the product has been discontinued. I replaced it with an M-Audio M-Track that is a USB device and had advertized zero latency audio monitoring on the side of the box.
I also installed a 4 port FW 800 hub to connect my drives to the single FW 800 port on my MBP. It's all the same pipe, I know, but at least now I can power off drives that I'm not using without having to mess with cables.
So some of the more immediate issues I was having are resolved, but that doesn't eliminate the need for scalable storage.
Why build vs. buy?
Educational. I've been wanting to learn more about managed storage for years. No better way than to build something from scratch.
Value. I think I can end up with a faster, higher capacity system than I could get elsewhere for the cost.
I like the idea of having a good indoor project to tinker with going into the fall/winter, this seemed like a very practical choice.
My career in high tech started long ago on the assembly line at Apollo Computer, but I've always liked building, tinkering with and fixing hardware. I remember the old Commodore 64 1571 5.25 disk drive hack that allowed you to put toggle switches on the back of the drive to change the addresses so that you could have two of them connected at once.
As I dig into this, the real concern is becoming the drives. The drives that you typically get in a consumer/retail setting are often vastly inferior in quality than something that a storage vendor like EMC would use. That's understandable, but building something reliable is key and having talked to disk drive qualification engineers at a couple of managed storage vendors, the horror stories about the high end product they work with has me worried that the consumer stuff that I use is reliable at all... Then again, that's what RAID is really all about; data redundancy and being able to replace drives on the fly if they fail. Still, I'm finding I'll have to pay a fair bit more per drive to get the sort of quality, speed and reliability that I'm after. I do plan to use my current collection of standalone drives to back up individual projects as they're completed.
So I'm tired of the pile of drives that constitute the storage for my edit suite. It's simply not scaling. I have a total of 8 TB across my firewire 800 drives. I'm typically using only one project drive at a time, but it's a pain in the ass to have to cable up just the ones I'll need for the nights work. I've noticed that the more I chain together, the slower response I get through my editing software - even if only one drive in the chain is active/in use.
After having spent the week having SAN's and iSCSI demystified in training classes at Dell/Equallogic, I was inspired to look into some of the free SAN/NAS solutions out there. I'm trying now to decide between openfiler and Free NAS. It looks as though my MBP should support the initiator software (from globalSAN) and I can use the one gig hard wired port for dedicated SAN traffic and wifi for all other traffic.
I looked around quite a bit and the consumer stuff doesn't seem to have the performance and the higher end stuff is way too pricey. The reviews on openfiler and Free NAS suggest solid reliability and excellent performance when configured properly.
It's looking like I can build a really awesome ten bay server with open filer for around $1000, maybe a little more depending on what size and type drives I decide to use. Should be interesting.