-
Guide: Planning an Unraid build — real usable capacity (TB→TiB + parity), transfer bottlenecks, and running cost
valid again, more work completed on the tools. Thanks again for the feedback.
-
Guide: Planning an Unraid build — real usable capacity (TB→TiB + parity), transfer bottlenecks, and running cost
Thanks trurl, apprecaite the feedback, suggestions adopted and hopefully the other unraid tools are more accurate as well
-
Guide: Planning an Unraid build — real usable capacity (TB→TiB + parity), transfer bottlenecks, and running cost
Good call - your right, appreciate the advice. I've updated the tool to include Unraid as its own storage type, with write speed modelled around the parity disk ceiling rather than treating it like a striped RAID array. Read behaviour is also handled separately now. At the risk of asking too much, would you mind taking another look and letting me know if the updated model gives a better representation of real-world Unraid performance?
-
-
Guide: Planning an Unraid build — real usable capacity (TB→TiB + parity), transfer bottlenecks, and running cost
Hi all, here’s a practical planning guide for Unraid builds and upgrades. It focuses on three things that come up constantly. Why your raw drive size never matches what Unraid reports How parity affects usable space when you are planning a build Why a faster network often does not translate into faster transfers I also put the calculations into a free tools page if you would rather plug in numbers than do it manually: https://needtoknowit.com.au/tools/ Capacity reality check: TB vs TiB Drive labels use decimal terabytes, but operating systems report tebibytes. That means every drive looks smaller before parity and before anything else. A common rule of thumb is advertised TB multiplied by about 0.909 to estimate what you will see as TiB. Parity planning: the simple mental model Unraid parity is not RAID5 or RAID6 striping, but for capacity planning the logic is still simple. If you run 1 parity disk, plan to give up roughly one disk worth of capacity. If you run 2 parity disks, plan to give up roughly two disks worth of capacity. Also remember that parity must be at least as large as the largest data disk. You can mix data disk sizes, but parity sets the ceiling. What people commonly forget to account for is the TB to TiB conversion, plus the space that gets used up by Docker images, appdata, VM images, backups, and metadata. It is also worth planning some headroom so the array does not live at 95 percent full. If you want an interactive breakdown that layers TB to TiB plus parity plus optional deductions and headroom, this calculator includes an Unraid option: https://needtoknowit.com.au/tools/raid-calculator/ Transfer speeds: why the upgrade did not fix it Transfers are limited by the slowest part of the chain. Common bottlenecks are 1GbE networking, spinning disks (especially with lots of small files), cache drive limits, the speed of the source machine, and workload differences (large sequential copies vs lots of small files). If you want a quick estimator that helps you see the likely limiter and gives a realistic transfer range: https://needtoknowit.com.au/tools/transfer-speed-estimator/ Running cost (optional) If you are trying to decide whether a build is worth it long term, power cost can be useful. This one defaults to Australian state averages, but you can enter any rate: https://needtoknowit.com.au/tools/power-calculator/ If there is a planning question you see newcomers ask repeatedly (capacity, cache sizing, mover behaviour, “why is it slow”), reply with the scenario and I can expand this guide with a worked example section.
TheEbbandFlow
Members
-
Joined
-
Last visited