Sanity check - 1 server vs 2 servers - Lemmy.World
lemmy.world
Hello! I’ve been running unRAID for about two years now, and recently had a thought to use some spare parts and separate my server into two based on use. The server was used for personal photos, videos, documents, general storage, projects, AI art, media, multitude of docker containers, etc. But I was thinking, it’s a bit wasteful to run parts that I use once or twice a week or less 24/7, there is just no need for the power use and wear and tear on the components. So I was thinking to separate this into a server for storage of photos, videos and documents powered on when needed, and then a second server for the media which can be accessed 24/7. Server 1 (photos, videos, documents, AI experiments): 1 x 16TB parity, 2 x 14TB array. I7 6700k, 16GB ram Server 2 (media, docker): 1 x 10TB parity, 1 x 10TB and 2 x 6TB array. Cheap 2 core skylake CPU from spare parts, 8GB ram. With some testing, server 2 only pulls about 10w while streaming media locally, which is a huge drop from the 90+ watts at idle that it was running when I had everything combined. I was hoping to use an old laptop I have laying around for the second server instead, which has an 8 core CPU, 16GB ram, and runs at 5w idle. I have a little NVMe to SATA adapter that works well but the trouble is powering the drives reliably. Anyways, pros of separating it out, lower power usage, less wear and tear on HDDs so I will have to replace them less frequently. Cons, running and managing two servers. Ideally, I’d like to run server 1 on the cheap 2 core skylake CPU (it’s only serving some files after all), server 2 on the laptop with 8 cores (but still have the issue of powering the drives), and then take the i7 6700 for a spare gaming PC for family. Alternative would be to just combine everything back into one server and manage the shares better, have drives online only when needed, etc. But I had issues with this, and would sometimes log into the web ui to find all drives spun up even though nothing was being accessed. Anyways, I hope all of that makes sense. Any insight or thoughts would be appreciated!

Hello!

I’ve been running unRAID for about two years now, and recently had a thought to use some spare parts and separate my server into two based on use. The server was used for personal photos, videos, documents, general storage, projects, AI art, media, multitude of docker containers, etc. But I was thinking, it’s a bit wasteful to run parts that I use once or twice a week or less 24/7, there is just no need for the power use and wear and tear on the components. So I was thinking to separate this into a server for storage of photos, videos and documents powered on when needed, and then a second server for the media which can be accessed 24/7.

Server 1 (photos, videos, documents, AI experiments): 1 x 16TB parity, 2 x 14TB array. I7 6700k, 16GB ram

Server 2 (media, docker): 1 x 10TB parity, 1 x 10TB and 2 x 6TB array. Cheap 2 core skylake CPU from spare parts, 8GB ram.

With some testing, server 2 only pulls about 10w while streaming media locally, which is a huge drop from the 90+ watts at idle that it was running when I had everything combined.

I was hoping to use an old laptop I have laying around for the second server instead, which has an 8 core CPU, 16GB ram, and runs at 5w idle. I have a little NVMe to SATA adapter that works well but the trouble is powering the drives reliably.

Anyways, pros of separating it out, lower power usage, less wear and tear on HDDs so I will have to replace them less frequently.

Cons, running and managing two servers.

Ideally, I’d like to run server 1 on the cheap 2 core skylake CPU (it’s only serving some files after all), server 2 on the laptop with 8 cores (but still have the issue of powering the drives), and then take the i7 6700 for a spare gaming PC for family.

Alternative would be to just combine everything back into one server and manage the shares better, have drives online only when needed, etc. But I had issues with this, and would sometimes log into the web ui to find all drives spun up even though nothing was being accessed.

Anyways, I hope all of that makes sense. Any insight or thoughts would be appreciated!