This paper presents the limitations of RAID 6 on today's large capacity SATA and SAS drives and proposes a new RAID 6 data protection strategy: increasing the number of parity bits or limiting drive capacities.
This whitepaper examines why traditional RAID 6 becomes less reliable when large‑capacity SATA and SAS drives are used in mainframe enterprise storage. It shows how increased drive capacity and higher Uncorrectable Bit Error Rates (UBER) reduce Mean Time To Data Loss (MTTDL).
Oracle proposes several alternatives, including increasing the number of parities to three (RAID Z3), limiting usable drive capacity, and employing globally shared hot‑spare drives. The paper presents calculations and graphs that illustrate the MTTDL improvements of these strategies.
| FC HDD Capacity | 146 GB |
|---|---|
| SATA HDD Capacity | 1 TB |
| SAS HDD Capacity | 2 TB |
| RAID 6 Drives per Group | 12 drives |
| MTBF (all drive types) | 1,000,000 hours |
Larger capacities increase the Uncorrectable Bit Error Rate and lengthen rebuild times, which dramatically lowers the Mean Time To Data Loss for RAID 6 groups.
Limiting usable capacity to about 85 % reduces the stress on the drive, lowering error rates and increasing MTTDL by roughly 62 % in Oracle’s tests.
A globally shared hot‑spare drive increases MTTDL by about 18 % compared with dedicated spares, improving overall RAID reliability.