Partitioned Data Storage Lockup Error Model

SWITCH TO API

Description

The Partitioned Data Storage Lockup Error Model simulates transient lockup conditions in spacecraft data storage units. Storage devices can experience temporary lockup states due to radiation-induced latch-up, firmware errors, or bus contention issues that prevent write operations. This model applies a configurable probability to lock individual storage units at each simulation step, temporarily preventing data writes. Locked units are automatically cleared at the start of each update cycle, simulating transient rather than permanent failures.

---

Example Use Cases

• Latch-up Analysis: Evaluate the impact of radiation-induced single-event latch-up on storage system availability.
• Write Availability Studies: Analyze the probability of successful data writes under intermittent lockup conditions.
• Fault Tolerance Testing: Validate that data handling software correctly manages unavailable storage partitions.
• Buffer Overflow Risk: Assess the risk of data loss when storage units become temporarily unavailable during high data rate periods.

---

Module Implementation

The Partitioned Data Storage Lockup Error Model is a Universe Model that attaches to a Partitioned Data Storage component and introduces transient lockup states on storage units.

Lockup Mechanism

At each simulation time step, the model evaluates each storage unit independently. The lockup process follows two steps:

1. Clear Previous Lock: Any existing lock on the unit is released
2. Apply New Lock: A new lock is applied with probability $p_{\text{lock}}$

This results in a transient lockup behavior where units may be locked for a single time step before being released.

Lockup Probability

For each storage unit, the probability of entering a locked state during a given time step is:

$$P(\text{lock})=p_{\text{lock}}$$

where $p_{\text{lock}}$ is the configured LockupProbability parameter. The default value of $10^{-6}$ represents a low but non-negligible probability suitable for modeling occasional transient faults.

System-Level Availability

For a storage system with $N$ units, the expected number of locked units at any time step is:

$$E[\text{locked units}]=N\cdot p_{\text{lock}}$$

The probability that all units are available is:

$$P(\text{all available})=(1-p_{\text{lock}}{)}^N$$

Effect on Storage Operations

When a storage unit is locked:

• Write operations to that unit are blocked
• Read operations may still be permitted depending on implementation
• Data already stored in the unit remains intact

---

Assumptions/Limitations

• Lockup states are transient and last only one simulation time step; persistent lockups are not modeled.
• All storage units have the same lockup probability; unit-specific failure rates are not supported.
• Lockup events are independent between units and between time steps.
• The model does not simulate the physical cause of lockup such as current surges or thermal effects.
• Recovery from lockup is automatic; manual intervention or power cycling is