RAID Comparison (RAID 6 Vs RAID DP)
RAID 6
“P”是一個奇偶校驗位,使用XOR校驗算法生成和“Q”是使用裏德所羅門糾錯碼生成算法計算
RAID DP
Where P denotes a Parity disk and DP denotes the Double Parity disk
其中P代表一個奇偶盤和DP表示的雙奇偶校驗盤
|
Feature |
RAID 6 |
RAID-DP |
|
Double Disk Protection 雙硬盤保護 |
Yes 是 |
Yes 是 |
|
Algorithm 校驗方法 |
SNIA RAID 6– XOR/ECC Evolved from RAID DP XOR/ECC校驗，從raid-dp演變而來 |
NetApp – Patented RAID-DP 有專利的raid-dp |
|
Algorithm Overhead 校驗方法帶來的性能消耗 |
High on Controller – As compute parity is more in comparison with RAID DP 耗能高 |
Low on Controller – As compute parity is direct XOR and Diagonal XOR 耗能低 直接校驗和對角校驗 |
|
Reconstruction Overhead* 重建帶來的性能消耗 |
On disk failure, rebuilt time is much longer Ø 6 Hours for single disk Ø 12 Hours for double disk failures 磁盤失效，重建時間長 Ø 單盤重建6小時 Ø 雙盤重建12小時 |
On disk failure, rebuilt times are faster and acceptable Ø 1.5 Hours for Single disk Ø 4 Hours for Double disk failures 磁盤失效，重建時間快可以接受 Ø 單盤重建1.5小時 Ø 雙盤重建4小手 |
|
Scalability 可擴展性 |
Limited. Recommended no RAID group scalability 受限制的，沒有raid組擴展的推薦 |
Scalable RAID group. Recommended to add 4 disks as best practices 可升級的raid組. 最佳實踐推薦增加4塊磁盤 |
|
Random write Performance 隨機寫性能 |
Regular XOR and ECC causes performance impact for Random writes 定期的XOR和ECC導致寫性能差 |
No performance impact for Random or sequential work loads 隨機和順序寫沒有性能影響 |
|
Vendor Best practices 供應商最佳實踐 |
Not recommended as a best practices by majority of vendors. RAID 1/0 or 5 沒有最佳實踐 |
Recommended by NetApp as the Best practices and defacto standard. 有推薦的最佳實踐 |
|
Parity Capacity Overheads 相同容量消耗 |
Two disks equivalent 消耗2塊磁盤容量 |
Two disks equivalent 消耗2塊磁盤容量 |
* Note – All timing mentioned are either through experience of scenarios or some technical document verification in general for 300GB FC. Subjected to change based on various controllers, OS version and capacity of the disks