Storage Infrastructure

Enterprise Backup Storage Infrastructure

Integrity checks at the storage layer.

ShadowCradle uses a purpose-built enterprise storage engine. Checksummed data integrity, automatic corruption healing, immutable snapshots, efficient replication, and deduplication - all built in.

Dedup

Identical blocks stored once

LZ4

Fast compression

15 Min

Min. snapshot interval

Auto

Corruption healing

Why Purpose-Built Storage?

Our enterprise storage engine was purpose-built for data integrity and high-performance backup. It is designed to detect and heal silent data corruption.

Other backup solutions layer their own file system on top of commodity storage. ShadowCradle's storage layer is built from the ground up - meaning every feature (replication, snapshots, deduplication) is built on a foundation designed for exactly this purpose.

Data Integrity Checksums

Every block verified with checksums. Corruption is detected and, where redundancy allows, repaired automatically.

Atomic Snapshots

Consistent point-in-time backups every hour or more. Snapshots are copy-on-write and do not copy data when taken.

Built-In Replication

Native replication powers ShadowCradle's multi-site sync. Only changed blocks replicate - no full copy needed.

Automatic Healing

Self-healing storage automatically detects and repairs corruption as it finds it. Single or multiple drive failures handled automatically.

Thin Provisioning

Only allocate storage as data fills. 100TB of thin-provisioned space costs far less than 100TB physical allocation.

Deduplication & Compression

Identical blocks stored once. LZ4 compression reduces storage with minimal CPU impact. Savings depend on your data.

How the Storage Engine Works

Every backup write passes through a purpose-built pipeline optimized for integrity and efficiency.

Incoming Backup Data

Block-level changes arrive from agents across all protected machines

Deduplication

Identical blocks across all jobs and machines are identified and stored only once

Space savings

LZ4 Compression

Unique blocks compressed with minimal CPU overhead before writing

More savings

Integrity Checksum

Cryptographic checksum recorded per block - corruption detected and auto-healed on every read

Auto-healing

AES-256 Encryption

Data encrypted at rest before any write to physical storage

Redundant RAID Pool

Written across Mirror, Single Parity, or Double Parity drives - survives 1–2 simultaneous failures

Atomic Point-in-Time Snapshot

Consistent recovery point captured and indexed for file, database, or VM restore

15-min minimum

RAID Resilience

RAID-1 (Mirror)

Two drives hold identical data. Single drive failure - automatic recovery. Simplest, highest performance.

Advanced RAID (Single Parity)

Advanced parity RAID without the write-hole vulnerability. Single drive failure protected. Efficient storage.

Advanced RAID (Double Parity)

Two simultaneous drive failures protected. Enterprise reliability for the most critical data.

The Setup Wizard Knows Your Drives

On first boot the appliance scans every installed drive, identifies its type and speed, and tells you exactly how to use it for peak performance. No guessing, no documentation hunting.

Data Pool

HDD or SSD

Your core protected storage. Drives form a redundant pool with your chosen parity level. The wizard recommends the best pool layout for the drives it detects.

Mix drive sizes - usable capacity is based on your smallest drives

Write Acceleration Journal

Fast SSD recommended

A dedicated fast drive that absorbs synchronous writes, then flushes to the data pool in the background. Helps reduce write latency spikes under heavy load.

Only 16–32 GB needed - perfect use for a small NVMe or leftover SSD

Intelligent Read Cache

SSD recommended

A dedicated SSD layer that automatically caches your most-accessed data. Hot reads are served from SSD without ever touching spinning disks.

Aim for 5–10% of your data pool size for best cache hit ratio

Metadata Accelerator

NVMe recommended

Small-block and metadata I/O is redirected to your fastest drive. Critical for environments with many small files or mixed-size workloads.

Even a small NVMe (64–128 GB) dramatically improves small file performance

Parity Advisor

As you add drives to the wizard, it updates its recommendation. You see protection level, usable capacity estimate, and failure tolerance before you commit.

2 drives

Mirror

50% usable1 drive failure

Simplest - both drives hold identical data

3 – 5 drives

Single Parity

67 – 80% usable1 drive failure

Good balance of protection and capacity

6 – 11 drives

Double Parity

67 – 83% usable2 simultaneous failures

Survives a full rebuild plus another failure

12+ drives

Double Parity (split pools)

83 – 92% usable2 failures per pool

Wizard suggests splitting into multiple pools for rebuild speed

What the wizard actually tells you

When you install drives and open the setup wizard, you see something like: "We detected 1x 256 GB NVMe and 4x 4 TB HDD. We recommend using the NVMe as your Write Acceleration Journal and Read Cache, and the 4 HDDs in Double Parity mode giving you 8 TB of protected storage. This protects against any single drive failure." You can override any recommendation.

Storage Performance Estimator

Configure your drives and see estimated capacity and performance. Toggle acceleration options to see their impact.

Drive Configuration

Number of Drives4

216

Drive Capacity

Drive Type

Parity→ single

Acceleration Options

Write Acceleration Journal

Dedicated fast drive for sync writes

Hot Spare Drives

Idle drives for automatic rebuild

Intelligent Read Cache

SSD layer for hot data reads

Capacity

Usable Protected

12.0 TB

Total Raw

16.0 TB

Failure Tolerance

Up to 1 drive

Parity Mode

Single

vdev Stripe Width Not Optimal

Your current 4-drive RAIDZ1 has a data stripe width of 3 - not a power of 2. ZFS performs best when the data portion of a vdev is 2, 4, 8, or 16 drives, as stripe segments align cleanly with storage blocks.

Pool Resilver After One Drive Failure

~25.0 hrs

to rebuild the failed drive

Reads 3 surviving drives × 4.0 TB at ~100 MB/s (throttled). Assumes 75% pool fill.

Moderate window. Avoid scrubs during resilver. Monitor SMART data closely and replace the failed drive immediately.

Hot Spare Drives

Without a hot spare, a drive failure leaves the pool in a degraded state until you physically source a replacement drive, insert it, and trigger a rebuild - typically 4–48+ hours depending on your procurement process. During that window, any additional drive failure risks data loss.

Enable hot spare drives to have 1 standby drive immediately available for automatic failover.

Estimated Performance

Sequential Read720 MB/s

Sequential Write470 MB/s

Random Read IOPS600 IOPS

Sync Write Latency

Without journal

8 ms

* Estimate based on the numbers you enter and assumed typical drive specs, not ShadowCradle measurements. Actual performance varies by workload and hardware.