Make NAS ready for 8k video editing

  • hi guys,


    I currently use the following:


    OS: OMV5


    System NVME: 1TB WD SN750


    Storage HDDs: 6x 4TB Seagate Ironwolf NAS ST4000 als ZFS RAIDZ1 (RAID5)


    64GB ECC RAM


    1GBit LAN


    So far I just used this NAS for some media storage/streaming and docker/crypto stuff. Now I became a video editor and my client is providing 8K VR footage to work with. I saw other editors editing their stuff directly on/over qnap QuTS machines with 10gbit ethernet and SSD cache. I have some question for my next upgrade:


    1. I plan to go from 6 to 8 HDDs. I have read one should use multiple VDEVs to keep the amount of HDDs/VDEV small and therefore gain higher IOPS. I think higher IOPS might be a thing when it comes to rendering since it de/encodes frame by frame. But how do I make use of multiple vdevs when having 8 HDDs? I have read 4 HDDs in one raidz1 is bad because of sector/blocksize?! Should I go for one raidz2 VDEV instead?


    2. Caching: I dont use SSD caching so far. I have read before I consider a cache I should max out RAM first. I did now and dont have performance issues so far so I did not bother until now. These QNAP QuTS youtubers use mostly 1-2 512GB SATA SSDs for caching. Does this make sense when it comes to higher IOPS, faster scrubbing in the timeline or faster rendering speeds?


    3. I definetly need a LAN upgrade. I have seen one youtube editing 4k footage from his NAS showing that the LAN througput was around 80MB/s. Since I work with up to 8K footage I would multiply this by 4 and have 320MB/s neccessary. I guess 2.5gbit LAN is close but enough?!


    Appreciate any experienced 4k/8k editors that share their knowledge

  • I can't say I'm a 4k/8k editor using a OMV as a server, but I am a film and video post production house in house system engineer/technician, looking after a lot if systems using shared storage.


    I can't give you OMV specifics regarding your desired setup, but I will throw out a few points to consider.


    1. simply saying 4K or 8K video means nothing other than the video frame size. The important information to make bandwidth and network decisions on is what it the bitrate of the files.

    Just picking numbers as an example, if a 1 minute 4k video and a 1 minute 8k video (or any resolution for that matter) is encoded at the same 400Mbps bitrate, the files will all be the same size and require the same network speed, drive speed and almost the same processing power since it is the same amount of data per second. Since bitrate dictates bandwidth and files size, a 1 Gigabit connection as an example has a maximum bitrate of 1000 Mbps, or 125 MBps (B=bytes, b=bits, there are 8 bits in a byte, so 1000/8=125). This also means that in that 1Gbe connection, if the file had a bitrate of 250Mbps and you could reach that maximum bandwidth, you could simutaneously transfer 4 files. However, you will never reach the theoretical maximum speed. The OS, systems hardware, drive speed and configuration, network cards and switches, and connection protocol (SMB/NFS/iSCSI), and even file system and fragmentation, will all play a part in determining the maximum speed you can achieve.


    2. If your drives are traditional spinning platter drives, while that use a sata-3 connection, they don't achieve the full bandwidth of the 6Gbps connection. Real world tests show the regular ironwolf drives achieving closer to 1.4Gbps (about 1/4 the connection speed).


    3. I you are wanting to use an SSD cache setup, you would need to somehow manage what gets cached to the SSD's, have enough drives to over-saturate the combined SSD bandwidth so as to avoid starving the cache in order to benefit from them, and control how this is all presented to you you. When used like this the spinning drives can sustain a stream, but the SSD cache adds a better random access and muti-stream ability. To my knowledge, none of this is native to OMV, so unless you can cobble together some kind of intelligent cache I don't think OMV is the right solution for this setup. OMV is designed as a NAS (Network Attached Storage), not a SAN (Storage Area Network), which is what is generally used in larger facilities that require this complexity. The QNAP boxes you mention are designed to be somewhere between the NAS/SAN/DAS (Direct Attached Storage) options as they often have a couple of thunderbolt ports that can be used as an iSCSI connetion too to minimize protocol "chattyness"


    If I were you, a simpler solution would be what I would consider, unless you plan to jump ship for QNAP. That solution would be to use your OMV as an archive, but an active edit project would be done from Direct Attached Storage or internal storage. A single USB3 connected SSD will out-perform any NAS solution using 2.5Gbe, particularly if USB connection utilizes UASP. If your DAS setup is several SSD's in a JBOD or a RAID connected via thunderbolt or USB-C, they will out-perform a 10Gbe and maybe even 25Gbe depending on the number of SSD's, drive configuration and speed of the USB connections.

  • Hi, just wondering what progress you made with your implementation. I edit a lot of 4k/5k on a 4bay QNAP over 10G Rj45 connection. I'm using Samsung EVO SSDs, no raid or cache disks on NAS. My scratch/cache disk is a local NVME in the desktop. I ALWAYS use proxy footage until render time (FCPX and Resolve). Even if I felt my computer could handle 8k, using proxy is much more efficient for cutting, coloring, and fx speed. It also reduces the amount of resources needed for background render cache needed. If you're using Premiere Pro, I feel your pain because proxy takes a little more time and the way Pr throttles the timeline is annoying. FCPx won't slow halt your timeline, but rather drop frames or distort the image in real time if overloaded.


    I've been using OMV on a Raspberry Pi just to host rendered video for streaming to clients outside of my VPN. While I liked some features of the QNAP, I think OMV may be the next high-end built solution because of hardware flexibility and ease to replace parts. It seems that QNAP uses more CPU and RAM by default


    Anyway, interested in other's experience with editing and related docker usage on their OMV systems, especially with 10G NIC drivers.

    NAS Spec 👇

  • My wife does some video editing in Resolve and I am an astrophotographer (the data workload is surprisingly similar) and both of us use local Gen4 performance oriented NVMEs for scratch, cache and processing. The NAS is for storage and archive. The cost for network gear to provide acceptable performance far exceeds justification until you have a team of 3 or more and even then, a 2tb performance NVME is under $200. The other justification is projects that exceed what 1 or 2 NVME's can hold (4tb or more).

    Main-Srv: Xeon E5-2650 V4 32gb, OMV7 - 6.5.11-7-pve, Compose 7.0.3, Backup 7.0, Kernel 7.0.3, KVM 7.0.1, Resetperms 7.0, Sharerootfs 7.0-1, Wetty 7.0-1

    Mini-Srv: Intel N95 8gb, OMV6 - 6.2.16-20-pve, Compose 6.11.3, Kernel 6.4.9, MergFS 6.3.8, Resetperms 6.0.3, Sharerootfs 6.0.3-1, Wetty 6.0.7-1

    Inlaw's Srv: AMD A10-7800 16gb, OMV6 - 6.2.16-20-pve, Compose 6.11.3, Kernel 6.4.9, MergFS 6.3.8, Resetperms 6.0.3, Sharerootfs 6.0.3-1, Wetty 6.0.7-1

Participate now!

Don’t have an account yet? Register yourself now and be a part of our community!