You Asked for RAID 1, But We Give You Something Even Better: RAID 6 and Ceph Triplication Explained
Learn why InMotion Cloud's RAID 6 and Ceph triplication protect your data better than RAID 1 alone. Failure tolerance comparison included.
When prospective customers ask us "do you offer RAID 1?", our answer is yes and then some. InMotion Cloud's storage architecture exceeds RAID 1's single-drive protection at every layer. This article explains what the common RAID levels actually do, what InMotion Cloud deploys instead, and why the difference matters when a drive fails at 2 AM.
What RAID Actually Does (and What It Doesn't)
RAID (Redundant Array of Independent Disks) is a method for combining multiple physical drives so that data survives a hardware failure, improves read performance, or both. The tradeoff depends on which RAID level you choose.
RAID 0: Striping Without Redundancy
RAID 0 splits data across two or more drives to improve read and write speed. There is no redundancy. If one drive fails, all data on the array is lost. RAID 0 is a performance tool, not a protection tool.
RAID 1: Mirroring
RAID 1 writes identical data to two drives simultaneously. If one drive fails, the other contains a complete copy and the system keeps running. RAID 1 is straightforward and effective, which is why customers ask for it.
The limitation: RAID 1 tolerates exactly one drive failure. If both drives in a mirrored pair fail at the same time a real risk during a lengthy rebuild process data is lost. A second drive often fails under the read stress of rebuilding a failed partner.
RAID 5: Single Parity
RAID 5 distributes data and parity information across three or more drives. One drive can fail and the array keeps running; parity data allows the missing drive to be reconstructed. Like RAID 1, RAID 5 tolerates only one simultaneous drive failure. During a rebuild, the array runs degraded and a second failure is catastrophic.
Why InMotion Cloud Uses RAID 6 for Local Storage
InMotion Cloud deploys RAID 6 on local (direct-attached) storage in every host node. RAID 6 extends RAID 5 by adding a second independent parity block meaning two drives can fail simultaneously without data loss.
Why this matters in practice: Modern high-capacity drives take hours to rebuild. During that window, a second drive failure destroys a RAID 5 or RAID 1 array. RAID 6 keeps the array intact even if a second failure occurs during the rebuild. The system continues operating in a degraded-but-safe state until both failed drives are replaced.
RAID 6 requires a minimum of four drives and dedicates the equivalent of two drives to parity. That overhead is the cost of genuine dual-drive redundancy and it is a cost InMotion Cloud absorbs so customers do not have to manage it themselves.
Network Storage: Ceph Triplication
For network-attached storage, InMotion Cloud uses Ceph, an open-source distributed storage platform. Ceph stores three complete copies of every object, each on a separate physical node (server) in the cluster. This is called triplication or 3x replication.
The key distinction from RAID: RAID protects against drive failures within a single server. Ceph triplication protects against entire server or node failures. If a host node loses power, goes offline for maintenance, or suffers a hardware failure at the server level, the other two copies remain fully available on separate nodes.
Ceph also self-heals. When it detects a lost copy whether from a failed drive or a failed node it automatically recreates the missing replica on healthy hardware. The process is distributed and runs in the background without requiring manual intervention.
Comparison: RAID 1 vs. RAID 6 vs. Ceph Triplication
The table below shows how each approach handles the scenarios that matter most.
| Feature | RAID 1 | RAID 6 | Ceph Triplication |
|---|---|---|---|
| Copies / Redundancy | 2 drives (1 mirror) | Dual parity across 4+ drives | 3 full copies across 3+ nodes |
| Drive failures tolerated | 1 | 2 simultaneous | Multiple (per node) |
| Node/server failure tolerated | No | No | Yes |
| Rebuild risk | High (1 failure during rebuild = data loss) | Low (tolerates a second failure during rebuild) | Minimal (automatic re-replication from healthy nodes) |
| Protection scope | Single server | Single server | Across servers and availability zones |
| Self-healing | No (manual replacement required) | No (manual replacement required) | Yes (automatic) |
| Typical use case | Basic server mirroring | High-value local storage | Distributed cloud storage |
InMotion Cloud's architecture addresses both rows where RAID 1 and RAID 6 show "No": server-level failures and self-healing. Ceph handles those cases at the network storage layer.
How the Two Layers Work Together
InMotion Cloud applies both technologies at the appropriate layer:
- Local (instance) storage uses RAID 6 protecting against up to two simultaneous drive failures within the host node.
- Network (volume and object) storage uses Ceph triplication protecting against drive failures, node failures, and network partitions with automatic recovery.
Most workloads on InMotion Cloud use network storage for persistent data volumes. A virtual machine's (instance's) root disk may reside on local RAID 6 storage, while databases, media files, and application data live on Ceph-backed volumes that survive the loss of any single node.
In practice: If a physical drive fails in a RAID 6 array, your instance keeps running with no interruption. If an entire host node fails, your network-storage volumes are already replicated across two other nodes and remain available within seconds.
What This Means When Something Goes Wrong
A drive failure is not an "if" it is a "when." Enterprise drives have rated failure rates measured in annual percentage terms. A storage cluster with hundreds of drives will see drive failures routinely.
RAID 1 survives one failure. During the hours it takes to rebuild the array, a second failure destroys all data. That is a real-world risk that happens frequently enough to be a known problem in storage engineering.
RAID 6 absorbs two failures including a second failure during rebuild. That alone is a material improvement over RAID 1 for any system where data loss is unacceptable.
Ceph triplication operates above the drive level. A full server going offline does not require a rebuild in the traditional sense; Ceph simply reads from the two remaining copies and begins asynchronously replicating a new third copy onto healthy hardware. The recovery burden is distributed across the entire cluster rather than concentrated on a single replacement drive.
Why We Don't Offer "Just RAID 1"
Configuring individual RAID 1 pairs for each customer would mean giving customers less protection than our standard infrastructure provides. Our decision to deploy RAID 6 and Ceph at the platform level means every customer benefits from dual-parity and triplication automatically without having to request it, configure it, or pay extra for it.
This is an architecture decision, not a marketing differentiator. We built InMotion Cloud's storage layer to match how modern distributed systems handle failure: assume drives and nodes will fail, design so that failure is invisible to the application, and automate recovery.
Next Steps
If you want to understand how your specific workload maps to InMotion Cloud's storage layers, the InMotion Cloud support team can walk you through which storage type fits your use case and how data is protected end to end.
For teams evaluating cloud providers specifically on storage reliability, ask any provider these three questions: how many simultaneous drive failures does your local storage tolerate, does your network storage survive a full node failure, and is recovery automatic or manual? The answers reveal how much redundancy is actually built in versus assumed.
To get started with InMotion Cloud, visit inmotionhosting.com/cloud or contact the sales team for a storage architecture review.