Comment on Realtek's $10 tiny 10GbE network adapter is coming to motherboards later this year
sugar_in_your_tea@sh.itjust.works 6 days agoAre you saying you’d be loading all that data strictly over the network instead of having a local copy that gets synced periodically? That would be terrible on a 100 mbit/s line… If that was my workflow, I’d run 10 gbit/s cable everywhere and make sure clients had at least 2.5G.
I use my NAS for local backups and streaming when we watch something as a family. 100 mbit/s would be fine for that use case.
SirEDCaLot@lemmy.today 4 days ago
Yes I am, and that is exactly the point. I do not want spinning disks in my desktop, or anyone’s desktop or laptop. Give the actual computer a fast SSD for the OS and programs, then store the big data on a NAS or server. How’s the computer access it from that server in real time.
At 100 megabits (10 megabytes per second) that isn’t very fun. Gigabit ethernet is 100 megabytes per second give or take. That is where it starts to become useful for storage, as most spinning disks themselves have a transfer rate between 100 and 150 megabytes per second.
But as you just pointed out, that can become a bottleneck. Especially if you have multiple people accessing the server. How much of a problem it becomes depends on what they’re doing. IE, 10 people editing photos can happily share a gigabit link to the server because they load the photo once and then the link sits idle while they work as the photo is cached in RAM, 10 people editing uncompressed high definition video will probably want a constant full gigabit to each of them because they’ll be using almost all of it constantly so you need a gigabit to each desk and 10 gig to the server (and a storage array with sufficient bandwidth)
sugar_in_your_tea@sh.itjust.works 4 days ago
You could look into automatic local caching for diles you’re planning to seed, and stick that on an SSD. That way you don’t hammer the HDDs in the NAS and still get the good feels of seeding. Then automatically delete files once they get to a certain seed rate or something and you’re golden.
How aggressive you go with this depends on your actual use case. Are you actually editing raw footage over the network while multiple other clients are streaming other stuff? Or are you just interested in having it be capable? What’s the budget?
But that sounds complicated. I’d personally rather just DIY it, that way you can put an SSD in there for cache and you get most of the benefits with a lot less cost, and you should be able to respond to issues with minimal changes (i.e. add more RAM or another caching drive).