Pick a RAID level, set how many disks you have and how big each one is, and the calculator draws where data and parity land on every disk. It shows the usable space in TB and in the TiB your operating system will report, how many disks can fail, how long a rebuild takes and how fast the array is, then explains the level in plain words and compares it with every other level for the same disks.
Table of Contents
How to read the disk diagram
Each column is one disk, drawn to scale. Blue blocks hold your data, orange blocks hold parity (the checksum used to rebuild a failed disk) and light blue blocks are mirror copies. Striped grey areas are space the array cannot use, and a grey disk is a hot spare that waits idle until another disk fails.
Change the size under any disk, remove one with the cross, or use quick fill to set them all at once. Click a row in the comparison table to switch to that level.
Choosing a level for a server
RAID 10 gives the best random write speed and the fastest rebuilds, which is why it is the usual choice for databases, virtual machines and busy hosting servers. RAID 6 or RAIDZ2 gives more usable space for backups and archives and survives two failed disks.
RAID 5 and RAIDZ1 on hard disks of 4 TB and larger are risky: a rebuild reads every other disk for many hours, and a second failure or unreadable sector in that time loses the array. On SSDs, where rebuilds are short, single parity is more reasonable.
Mixed disk sizes and Synology SHR
Classic RAID uses every disk only up to the size of the smallest one, so a 16 TB disk in an array of 8 TB disks wastes half its space. Synology Hybrid RAID (SHR) splits disks into layers and builds a separate RAID 5 or RAID 1 on each layer, so larger disks are used as long as at least two disks share the extra space. ZFS gets similar flexibility by combining mirrors of matching pairs.
Hardware RAID, mdadm or ZFS
Hardware controllers with battery- or flash-backed cache perform well but tie you to that controller model. Linux mdadm software RAID is portable and supported by the cPanel and DirectAdmin installers. ZFS, on Proxmox or TrueNAS, adds checksums that detect silent corruption, but it needs plenty of RAM and should not sit on top of a hardware RAID volume.
RAID Calculator at a glance
Official documentation: mdadm manual, OpenZFS documentation, Synology Knowledge Center.
Related guides: WHM Backups S3: Reliable Remote Backups and Test Restores · DirectAdmin Backups S3: Reliable Admin, Reseller and User Backups · Disk full on a production server: recovery runbook.
Frequently asked questions
Is RAID a backup?
No. RAID protects against a failed disk only. Deleted files, ransomware, a faulty controller or a fire affect every disk at once, so you still need backups on another server.
Can I mix disk sizes?
Yes, but classic RAID treats every disk as the size of the smallest one and the rest is wasted. Synology SHR and ZFS mirrors of matching pairs make better use of mixed disks.
Why is my usable space lower than the calculator shows?
File systems keep some space back (ext4 reserves 5% for root by default) and ZFS reserves about 3% plus metadata. The operating system also shows TiB, which is about 9% less than TB.
How long does a rebuild take?
Roughly the size of one disk divided by the rebuild speed: an idle 16 TB hard disk at about 120 MB/s takes around 37 hours, and a busy server takes longer. SSDs rebuild many times faster.