I would like to retire my Synology DS415play (3x4 TB Red) and my DELL OptiPlex SFF (Windows 10, PMS) with the 8 TB USB drive.
Ideally I would like to get an head less 19" system with at least 6 x 3,5" SATA Bays, to put in the 3x4 TB WD Red and to start with a 12 TB IronWolf for the Plex media. Having the OS (Windows or Linux) for the Server on a M2.PCIe 250 GB SSD should be enough. The system should run PMS, present the storage over the network and should be able to host VMs/Docker (Unifi Controller, Monitoring, Smart Home …).
DVR would be added later and would certainly increase storage requirements. I thought about getting a network enabled DVB-S2 receiver.
Performance wise the system should be able to deal with a single 4k stream + full HD streams to supported Hardware (LG TVs with WebOS Player, which seems to direct play 4k HDR streams).
For a case, a mainboard, the cpu, ram, the m2 ssd I wouldn’t spend over 1k US$ if possible.
The case can hold natively 11 drives, plus you can use a 2x5.25 -> 3x3.5 converter to give a total of 14 drives in that chassis, OR you could use 3x5.25 -> 4x3.5 converter for 15 drives if you don’t need a 5.25.
The power supply should handle all of that without a graphics card. The motherboard has onboard graphics for installation and setup purposes.
Plex is not a RAM heavy program, so only 16GB is necessary.
8 Threads and 16 cores is pretty decent for a chip at that price point, and the single threaded performance of that chip isn’t bad. Some may tell you to go Intel w/ graphics on chip to take advantage of the Hardware Acceleration, but I haven’t found a need for this yet (I have the 2700X in my system currently)
I never really considered doing this, as I generally don’t run into any sort of bottleneck transcoding. I’ve tried off of my NVMe and off of a single HDD, and didn’t notice any statistically relevant performance increase while transcoding. The limitation I typically run into while Transcoding is the Processor and GPU… Any other reasons to consider doing this besides a “faster transcode”? Which doesn’t seem to be the case in my situation??
(just as an example. Same movie/file. Using NVENC x264 I transcode a 1080p file at approximately 1200fps on both NVMe and on a SATA 6 HDD)
Sure, but you could do that just as easily by placing the transcode on the Storage RAID, right? with 5-6 drives running, it shouldn’t really be a factor I wouldn’t imagine
Edit: Personally speaking, I just use a 1TB Scratch drive, I also have my Windows Page file there, my scratch drive for Photoshop, Premiere, and a couple of other programs, as well as my temp transcoding directory for Plex
I know that there are a lot of tech heads out there, who love hardware science. But, as powerful and complex as PLEX is, it doesn’t require a whole lot! I have a 4+ year old HP with 8MB of conventional memory, hooked up to a 2TB USB drive, and a 8TB USB drive. I get bluray quality on my home network. But the reality is, none of that hardware stuff matters, if you’re streaming from a seedbox, or you’re accessing your home server (requiring the internet), and your bandwidth is s**ts. Regardless of how “powerful” your computer is, bad bandwidth is the killer of streaming experiences!
Additionally, if you want to use some sort of software RAID, you’d also need addtional RAM to use that efficiently. The bigger the array, the more RAM you’ll need.
You could do that as well but RAM is cheap and better performing. On Linux you don’t need to do anything but point the transcode directory to /dev/shm
For me, this was the primary reason for using a RAM Drive.
Unfortunately my RAID Controller on my HP ProLiant ML350 G6 is quite old, and doesn’t support TRIM, so I mirrored a pair of 128GB SSD’s together for the OS and removed the unnecessary wear by moving the Transcode area.
As I said, I am aware that a number of people have indeed worn out their SSD’s prematurely by not doing so.
Yes and no… Not sure how busy your server gets, but there are times when I may have 4 or 5 steams running.
I noticed that my RAID storage was getting VERY busy as it would go something like this…
READ file from Storage → Transcode and WRITE the transcoded chunks back to the Storage → READ the transcoded chunks from the Storage to stream to user.
Do this 4 or 5 times, and the storage gets unnecessarily busy.
Instead, with the Transcode area in RAM Drive, it goes like this…
READ file from Storage → Transcode and WRITE the transcoded chunks to the RAM Drive → READ the transcoded chunks from the RAM Drive to stream to user.
One question beside the hardware, which seems to be very controversial.
What OS to choose?
I thought about to pick “Unraid” - but how well maintained is it? How flexible is it to set transcode to the RAM? Does it use the full power of the CPU or is it virtualized?
