RAID and ZFS storage calculator

Choose the disks and the layout. See the usable space in TB and TiB, how many disks can fail, and what it means for speed.

Your disks

Several smaller groups are faster and rebuild sooner, but use more disks for parity.
Disks kept idle to take over at once when one fails.

ZFS RAID-Z2, 8 × 16 TB

Usable space93 TB84.6 TiB as the OS shows it
Disks that may fail2 diskswithout losing data
Space efficiency73%128 TB raw
  • data
  • parity
Layout
ZFS RAID-Z2
Sequential read
about 6 × one disk
Sequential write
about 6 × one disk
Random I/O (IOPS)
about one disk

ZFS figures leave out about 3% that ZFS keeps in reserve. Keep pools below about 80% full for good speed.

RAID is not a backup: it protects against a failed disk, not against deleted files, ransomware or a lost server.

Storage servers

Our storage servers come with 4 to 14 large hard disks plus NVMe for the system, ready for RAID 6 or RAID-Z2. Refurbished servers with several disks also turn up in the server auction.

Keep this result

Email me this result

One email with a link that opens this result again, so you can come back to it or share it with your team. Nothing else, unless you tick the box for news.

How the calculator works

Usable space is the number of data disks times the disk size. Mirrors keep one copy per pair, so half the disks hold data. RAID 5 and RAID-Z1 give one disk per group to parity, RAID 6 and RAID-Z2 two, RAID-Z3 three. For ZFS the result leaves out about 3% that ZFS keeps in reserve, so it matches what zfs list shows on a new pool.

Disk makers count in decimal terabytes (1 TB = 1012 bytes), while Linux tools such as df often show binary tebibytes (1 TiB = 240 bytes). That is why a 16 TB disk shows as about 14.6 TiB. The calculator shows both.

Which layout should I use?

  • RAID 1 or ZFS mirror for two system disks: simple, fast and the standard on every DediGermany server's NVMe.
  • RAID 10 or ZFS mirrors for databases and virtual machines: the best random speed and quick rebuilds, at half the raw space.
  • RAID 6 or RAID-Z2 for large hard disks: two disks can fail, which matters because rebuilding a 16 TB or 22 TB disk takes many hours.
  • RAID-Z3 for very wide groups (12 disks or more) or archives that must survive three failures.
  • RAID 0 only for scratch data you can recreate: one failed disk loses everything.

Speed figures are rules of thumb. Sequential throughput grows with the number of data disks; random I/O depends on the number of mirrors or RAID-Z groups, because each group behaves roughly like one disk for small random writes.

Questions

How much space do I lose with RAID-Z2?

Two disks per RAID-Z2 group hold parity. With 8 × 16 TB disks in one group you get about 6 × 16 = 96 TB, or about 93 TB after ZFS's reserve. Wider groups lose a smaller share to parity.

Is RAID 5 still safe with large disks?

It works, but a second disk failure during a rebuild loses the array, and rebuilds of 16 TB disks take a long time. For disks of 8 TB and more, RAID 6 or RAID-Z2 is the usual choice.

Why does my server show less space than the calculator?

Check whether it shows TiB instead of TB, and remember the file system's own overhead. A 16 TB disk is about 14.6 TiB; ZFS also keeps about 3% back.

Is RAID a backup?

No. RAID keeps the server running when a disk fails. It does not protect against deleted files, ransomware or a lost server. Keep a separate backup, for example our backup storage add-on in the same data centre.

Can I choose the RAID layout on a DediGermany server?

Yes. You get full root access, so you can set up mdadm RAID or ZFS as you like, for example from the rescue system. The NVMe system disks are delivered as RAID 1.

Not sure which server fits?

Tell us what you want to run. We reply around the clock, usually within the hour.

See all serversAsk our team