Recommendations for Storage Preparation

Recommendations for storage preparation with Oracle ASM are discussed in this topic.

The following are guidelines for preparing storage for use with Oracle ASM:

• Configure a separate disk group for the following:

  • Oracle Cluster Registry (OCR) and voting files

  • Grid Infrastructure Management Repository (GIMR) files

  • Database data files

  • Fast recovery area

• The number of LUNs (Oracle ASM disks) for each disk group should be at least equal to four times the number of active I/O paths. For example, if a disk group has two active I/O paths, then minimum of eight LUNs should be used. The LUNs should be of equal size and performance for each disk group.

  An I/O path is a distinct channel or connection between storage presenting LUNs and the server. An active I/O path is an I/O path in which the I/O load on a LUN is multiplexed through multipathing software.

• Ensure that all Oracle ASM disks in a disk group have similar storage performance and availability characteristics. In storage configurations with mixed speed drives, such as flash memory and hard disk drives (HDD), I/O performance is constrained by the slowest speed drive.

• Oracle ASM data distribution policy is capacity-based. Ensure that Oracle ASM disks in a disk group have the same capacity to maintain balance.

• Configure a minimum of three failure groups for normal redundancy disk groups and five failure groups for high redundancy disk groups to maintain the necessary number of copies of the Partner Status Table (PST) to ensure robustness with respect to storage hardware failures.

• External redundancy disk groups can be used when external hardware, such as a storage array, provides redundancy. High-end storage arrays generally provide hardware RAID protection. 

  Use Oracle ASM mirroring redundancy when not using hardware RAID, or when you need host-based volume management functionality, such as mirroring across storage systems. You can use Oracle ASM mirroring in configurations when mirroring between geographically-separated sites (extended clusters).

• Minimize I/O contention between Oracle ASM disks and other applications by dedicating disks in Oracle ASM disk groups.

• Choose a hardware RAID stripe size that is a power of 2 and less than or equal to the size of the Oracle ASM allocation unit.

• Use the Oracle ASM Filter Driver feature to provide consistent device naming and permission persistency.

找了些相关资料

这是目前，我认为比较恰当的解释。

oracle的建议仍然是每个ASM磁盘组至少有4个LUN。这样做的原因是在操作系统级别为存储提供了足够的IO队列。如果每个磁盘组只有一个LUN，则操作系统将仅创建一个IO队列。 IO队列通常最多可容纳256个命令，具体取决于您的操作系统和配置。对于强大的存储系统，这可能是瓶颈，而您的SSD存储可能是此选择。

像在oracle中一样，最好的方法是评估您的环境。如果您的存储供应商对LUN队列深度和配置有建议，请咨询您的存储供应商。如果他没有一个，则首先在存储上设置4个LUN，然后使用1-4个LUN对Orion进行io测试。

使用更多的LUN是否可以提高性能？除了纯粹的性能外，还要考虑的另一点是LUN的管理。在我的环境中，我使用至少4个LUN，并且通常使用更多个LUN，因为这可以使磁盘空间以有用的增量（例如1）增加。需要时为100-200 GB。当空间不足时，单个1 TB LUN将需要添加大小相等的1 TB LUN。