Galaxy Metal Gear 65 manual Changing the Read-Ahead Cache Size

Galaxy 65 User Guide

5If you selected Include Listed Hosts or Exclude Listed Hosts, you must create a host list.

All hosts already on an include or exclude this for the partition are listed in the Host Table for Array panel. You can change a host nickname or delete a host from the list.

To add a host that is already on the global host list, select the host in the Add Existing Host for Array panel and click Add Existing Host. For more information about the global host list, see 3.4.4.2, ”Creating Nicknames for Host WWNs”, on page 28.

To add a host that is not already on the global host list, enter the host WWN and nickname in the Add New Host for Array panel and click Add New Host.

3.4.5 Changing the Read-Ahead Cache Size

You can change the read-ahead cache size for each partition. This setting controls how much data the SM reads ahead and stores in its cache memory during sequential reads.

If you want to change this size, you should be prepared to monitor the system performance using the array statistics and adjust the size until you find the optimal size for your application.

The default setting is one chunk for the first access in a sequential read and one stripe for all subsequent accesses. The size of the chunk is based on the chunk size used when you created the array (default is 64 KB). The SMs treat volumes and mirrored arrays (RAID 1) internally as if they have a stripe size of 64 KB, even though they are not striped.

If you specify a read-ahead cache size, that amount of data is read first, and the same amount is read for all read-ahead accesses.

Read-ahead is triggered by two back-to-back accesses to consecutive logical block address (LBA) ranges. Read-head can be forward (that is, increasing LBAs) or reverse (that is, decreasing LBAs).

Setting the read-ahead size to 0 turns off read-ahead cache. This is useful if the host is triggering read- ahead for what are random accesses. This can happen if the host breaks up the random I/O into two smaller reads, triggering read-ahead. You can use the partition statistics read histogram to determine what size accesses the host is doing.

Setting the read-ahead size to a very large size (say four to five stripes) can help improve the performance of multiple (three or more) sequential read streams, in some cases improving performance by three times. This would be useful for providing multiple video streams, for example.

To change the read-ahead cache size:

1From SAM, select Manage Galaxy 65 Subsystem > LUN Management > Partition Menu > Read Ahead Cache.

The Read Ahead Cache page displays with a list of all existing arrays

2In the Select an Array to View LUN Information panel, select the array whose partition you want to assign a new read-ahead cache size.

3In the Partition Menu panel, select the partition name.

30