Shulou(Shulou.com)06/01 Report--

In view of what needs to be considered when selecting ASM for storage partition, this article introduces the corresponding analysis and solution in detail, hoping to help more partners who want to solve this problem to find a more simple and feasible method.

Raid 1, 0, 0, 1, 1, 5.

Raid 0room1, first configure raid1 on the disk group of raid 0, so that the underlying disk group is a striped configuration of raid 0, and raid1 can bring disk redundancy, but if one of the disks in the disk group is damaged, the whole raid 0room1 must restripe the disk. During the reconstruction striping period, the disk redundancy is also invalid, so the recovery time of the disk group is longer. But the performance will be better because raid 0 is faster in terms of read and write.

Raid 1 has striped configuration of raid 0 on the basis of raid 1. Raid 1 makes data redundancy for disk groups, which is a complete redundancy of disk bytes. If one disk is damaged, only one disk is damaged, because raid 1 is the underlying disk group configuration, so it will not cause disk damage, and the recovery time is just the time it takes to recover the disk. Since both disk writing and reading require extra Icano for disk redundancy, the performance is not as fast as that of raid 0mm1, but the recovery time is short.

Compared with raid1, raid 5 uses less disk capacity, has block-level redundancy, is not as redundant as raid1 + 0, and its performance is not as fast as raid1 + 0. It takes a little more time or twice the time to read and write a single disk or multiple disks. Consider using orion to simulate OLTP transactions and test them before the system goes online.

After raid is configured, file system is considered. Raw partition or ASM,ASM provides the same operability as managing file systems and provides the performance of raw partition. ASM generally only needs to use two disk groups for the database, one disk group for data access, one disk group for flashback, access to archives, backups, flashback logs, and so on. When we do LUN for disk partitions, we all start with the outermost cylinder of the disk, so the disk group used to access data can be divided by the cylinder of the first half of the disk, which has the fastest performance, and the cylinder of the second half can be used as a flashback disk group.

Starting with Oracle 10g R2, database files for both RAC and single instance databases can be stored in the same ASM instance for management.

Storage selection:

1.ASM mirror function, normal mode

2.RAID 1: 0 and ASM external

3.RAID 5 and ASM external

ASM can be divided into thick and thin AU,AU can be divided into 1m coarse policies. In the initial Data File request, 1m AU is allocated first. If it is a thin AU, then each ASM DISK is assigned the size 128K of 1 ASM DISK 8 of AU, so that each ASM DISK will be divided into 1 AU 8. Log files can be fine-grained. Log files and data files do not need to be divided into different disk groups and will not affect performance.

RAID 0 + ASM is the best way to stripe.

Now that we know how to select storage, let's take a look at a case of storage configuration. If you choose ASM striping:

A customer needs to do a 1T database with a 2T flashback area, and the disk is 73G type.

1T Data DG 16073G

2T Flashbackup Recovery DG 32mm 73G

Set up 8 Array, each Array = 12 * 73G, a total of 96 LUN, do raid 1, only 48 LUN

Why 48 LUN? We can calculate that if it is a 1T, it needs at least 14073Greb 2T flash at least 24073G, that is, at least 42 LUN, while to do a disk RAID1 mirror, 42LUN needs at least 86 LUN, accounting for half the image space. However, 86 cannot allocate LUN by 8 groups.

This LUN partition method is to use the whole disk as a LUN. In order to make use of the partition function of the disk, which is also the cylindrical partition function mentioned earlier, you can divide the disk partition to do LUN, and you can divide fast partition and slow partition to do LUN. This allocates all the data to the fast partition portion of the LUN, making the data cross all LUN.

If this case chooses RAID 0room1 for disk striping and mirroring, let's take a look at:

If you do raid 01g, then each Array can use 1273G, that is, 876G can be used as ASM disk group, there are a total of four 876G. At this point, we can create a large LUN based on 876G and use 4 LUN. ASM striping across all LUN,LUN and across all physical disks via RAID 0. Data is better distributed across physical disks. However, we use a larger LUN here than we did with the previous ASM+RAID1 configuration, and if we want to use fast partitions, we need to create an additional LUN, which ends up with 2 LUN. Hardware-level stripes try to choose 1m to match ASM's AU.

Summary:

1. Consider hardware-level protection.

two。 Try to choose the outer edge cylinder of the disk as LUN, which can provide higher performance.

3. Small disks, such as 73G, with higher rpm 15K. The spindle (disk) affects the positioning time and data transmission.

4. For higher bandwidth, maximize the number of spindles of the disk group.

This is the answer to the question about what you need to consider when choosing ASM for storage partition. I hope the above content can be of some help to you. If you still have a lot of doubts to be solved, you can follow the industry information channel for more related knowledge.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.