Storage Classes
IronShard offers three storage classes that let you choose the balance of cost, durability, and performance that best fits your data.
Each class maintains IronShard's GDPR-everywhere guarantee, and all classes use our adaptive redundancy system, dynamically adjusting how many copies or encoded fragments of your data exist and where they are placed, based on how often and how quickly you need to access them.
Unlike traditional object storage, where the entire bucket or object must fit a single fixed configuration, IronShard continuously optimizes storage within each object at the sub-object (chunk) level.
That means your data automatically becomes faster when it's hot and more efficient when it's cold, without any manual management or unpredictable pricing.
Storage Classes Overview
| Storage Class | Data fragments | Redundant fragments | Locality target for hot chunks | Repair priority | Eligible providers/regions |
|---|---|---|---|---|---|
| Budget | 4 or 6 | 1-2 (default 1) | none | low | cost-optimized set |
| Standard (default) | 6 or 8 | 1-4 (default 2) | 4 | medium | broad regional set |
| Performance | 6 or 8 | 2-5 (default 3) | 5 | high | full set, including hyperscalers |
Budget
For backups, archives, and cold datasets
The most cost-efficient multi-provider class. Ideal for personal use, proofs-of-concept, or data that is rarely accessed.
Your data remains safely distributed across independent providers, but redundancy and repair activity are minimized to reduce storage cost.
Standard (default)
For active projects and general workloads
A balanced class designed for ongoing workloads, development environments, and content that's frequently accessed but not latency-critical.
Adaptive redundancy automatically increases protection or locality when your data gets hot, then scales it back as it cools down, all within a predictable flat price.
Performance
For analytics, AI, and latency-sensitive data
Our fastest class, built for compute-intensive and interactive applications.
Performance class data benefits from predictive cache warming, low-latency locality placement, and straggler-smoothing techniques that minimize tail latency even under load.
You always get the best possible response times within the SLA, without needing to over-provision or manually move objects between tiers.
How IronShard's Classes Differ from Traditional Tiers
Conventional object storage systems (like AWS S3 or GCP Storage) use fixed storage classes and optional “intelligent tiering”, moving entire objects between static pools depending on how long they've been idle.
That approach changes latency and sometimes price when data moves between tiers.
IronShard takes a different path:
- We don't move objects between pools.
Instead, we adapt redundancy and placement within the same class, at the sub-object level. - You always pay a flat price per class.
Internal optimizations never affect your bill. - You never lose performance suddenly.
Hot regions of your data are proactively accelerated before you notice slowdowns.
At a Glance
| Traditional Tiering | IronShard Storage Classes |
|---|---|
| Whole-object lifecycle automation (cold objects moved to slower pools). | Continuous, intra-object optimization of redundancy and placement. |
| Fixed redundancy per class. | Adaptive redundancy tuned in real time to your data's activity. |
| Price and latency change when data is moved. | Flat price, steady SLA-grade performance within each class. |
Why This Matters
- Predictable pricing: Each storage class has a fixed logical price per TB. You can choose Budget, Standard, or Performance, and never worry about surprise charges when your data becomes active.
- Automatic optimization: IronShard continuously balances performance and cost behind the scenes, so you always get the right speed and reliability for your data's current usage.
- Consistent SLA: Your data stays within its class SLA for durability and latency, regardless of internal redundancy changes.
- Smarter sustainability: By matching protection to demand, IronShard reduces wasted storage and energy, helping you meet green-IT goals without compromising safety.
How It Compares
Hyperscaler Approach
Pick a storage class, we handle redundancy.
In large cloud systems, redundancy is fixed and permanently overbuilt to meet a global SLA.
Even idle data maintains full replication or erasure coding overhead until a lifecycle policy moves it to a colder, slower, and cheaper pool.
IronShard Approach
Pick a storage class once, we handle the rest.
We keep redundancy minimal when data is cold and scale it up automatically when it gets hot.
Each class defines a range of redundancy and performance levels, and the system adapts within that range continuously: no object rewriting, no tier migrations, and no billing surprises.
The Value for You
- Simple choices: Budget, Standard, or Performance, just pick what fits your workload.
- Always right-sized: Your data's protection and locality adjust automatically to how you use it.
- No surprise bills: The price per TB stays fixed inside each class.
- Better efficiency, faster access: Hot data streams faster; cold data stays affordable.
- GDPR-everywhere: All data, in every class, is stored only in GDPR-compliant regions and providers.
Choosing the Right Class
Use this quick guide to select the best class for your workload:
| Use Case | Recommended Class | Key Benefits |
|---|---|---|
| Personal backups, archives, logs, and cold datasets | Budget | Lowest cost per TB, ideal for long-term storage, automatic redundancy scaling. |
| Development environments, general-purpose apps, active project data | Standard (default) | Balanced cost and performance, adaptive optimization for frequently accessed data. |
| AI, analytics, real-time processing, low-latency applications | Performance | Predictive cache warming, best tail latency, optimized for compute proximity. |
Tip: You can always change the storage class later through lifecycle rules: IronShard will rebalance your data automatically without downtime.
IronShard gives you the simplicity of storage classes like AWS or GCP, but inside, it adapts automatically, so you always get the SLA you paid for, at lower cost and faster tail latency.
