More from every drive.
Storage software for flash and hard drives, on the hardware you choose.
One architecture.
From one server to petabytes.
Fast or efficient.
Pick one. Until now.
Mirroring
Fast, but wasteful.
Two full copies of everything: you buy every drive twice. Also called RAID‑10.
Parity
Efficient, but slow.
Fewer drives, but every small write costs extra reads and writes, and SSDs wear out early. Also called RAID‑5.
maxRAID
Fast and efficient.
The drive savings of parity, faster than mirroring in our tests, with compression on top.
Scattered writes in. Orderly writes out.
maxRAID collects small writes and stores them in large, even passes.
Faster than mirroring. Fewer drives.
Small random writes, on the same six SSDs.
Measured: 4K random writes per second (IOPS), six 3.75 TB NVMe SSDs, maxRAID‑5 (five data drives, one parity).
Fewer drives for the same capacity.
120 TB of usable storage on 15 TB SSDs.
Modeled. Typical data is 30% compressible; with highly compressible data, as few as 5 SSDs.
Five years. No SSD replacements.
What the SSDs cost over five years, under the same heavy‑write workload.
Modeled: 120 TB on 15 TB SSDs; maxRAID sized without counting on compression, drive life with typical data (30% compressible); mirroring = RAID‑10, parity = RAID‑5.
One block storage stack. Every scale.
Your volumes move between setups — and get live copies — without downtime.
Add nodes. Stay online.
Shared storage that stays online, from one server to many.
Web console and APIWorks with Proxmox VEBlock, file and S3 access
Every system. A single pane of glass.
Every maxRAIDOne system you run, in any site, with Grid and maxObject.
All systems, one viewNo firewall setupAll alerts, one place
Servers fail. Service doesn’t.
Built for cloud providers and AI, on your own hardware.
Pre‑release. A 10 × 10 grid: ten 8+2 arrays, each surviving any two failed drives or two whole servers.
Less flash for every usable terabyte.
1.8 PB usable on 15 TB SSDs. Both survive any two drive failures.
Modeled: maxRAID Grid with 10+2 arrays, against Ceph keeping three copies.
Add servers. Add performance.
Each volume spreads across arrays, with no central bottleneck.
Design targets, when the network can carry it. Operations = IOPS. NVMe‑oF (TCP, RDMA), GPUDirect, Kubernetes.
More terabytes to sell.
Same 1,000 drive slots. Same protection against any four drive losses.
Usable capacity, 1 PB = 1,000 TB. CMR and host‑managed SMR drives; maxObject 36+4L+3G LRC against Ceph RGW 12+4. Pre‑release: figures may change.
One disk per read.
Objects are stored together, so reading one usually touches a single disk.
Design arithmetic for an S3 GET: about 8 ms of drive time per small read, against 96 ms for Ceph RGW at 12+4.
Back to full protection in minutes.
Every drive takes a share of the rebuild. Customers keep reading and writing.
Design arithmetic: 50 servers, 1,199 drives sharing the rebuild. Only live data is rebuilt.
See it on your own hardware.
Try maxRAID free for 120 days, on arrays up to 100 TB. Talk to us about Grid and maxObject betas.
wildfire-storage.com
sales@wildfire-storage.com · +1 707 MAX‑RAID (629‑7243)
maxRAID
Software RAID for SSDs.
RAID‑6
with maxRAID
faster than RAID‑10.
Measured: mixed reads and writes (70/30, 4K) on the same six NVMe SSDs.
With fewer drives.
Modeled: 120 TB usable on 15 TB SSDs. RAID‑10 mirrors everything; maxRAID‑6 uses 8 data + 2 parity drives, or 6 + 2 when inline compression meets typical data (30% compressible).
For a third of the cost.
Modeled five‑year SSD cost: 120 TB usable, 15 TB SSDs at $4,410, the same heavy‑write workload. RAID‑10 drives replaced in year 3; maxRAID drives last the five years. Drives only.
maxRAID Grid
Block storage across many servers.
Lose a server. Keep serving.
Pre‑release. 10 servers × 10 SSDs: each logical array (8 data + 2 parity) has one drive in every server and survives any two failed drives or servers.
Less flash.
Modeled: 1.8 PB usable on 15 TB SSDs, both surviving any two drive failures — maxRAID Grid with 10+2 logical arrays, Ceph with three full copies.
Add servers. Add speed.
Design targets, when the network can carry it: up to 20 GB/s and 2 million operations per second per logical array.
maxObject
S3 object storage on hard drives.
Same drives.
35% more to sell.
Pre‑release. Usable capacity in the same 1,000 drive slots, protection against any four drive losses: maxObject 36+4L+3G against Ceph RGW 12+4.
One disk per read.
About 5× faster than Ceph.
Design arithmetic, not a benchmark: weighted by a week of real public‑cloud traffic, maxObject needs about one sixth of the drive time of Ceph RGW at 12+4. A small S3 read uses one disk instead of twelve.
Back in minutes.
Design arithmetic: 50 servers, 1,199 drives sharing the rebuild of a full 26 TB drive. Only live data is rebuilt.
Ask us. We’re right here.