Storage Processor Cache Test

To service I/O requests faster, to reduce disk overloads, and to eliminate disk abuse, the read/write caches should be sized with sufficient memory pages.

Cache page size determines the minimum amount of storage processor memory used to service a single I/O operation. Given below are some guidelines to right-size your cache:

• Default of 8KB is fine for majority of workloads.
• Default of 8KB is fine for majority of workloads.
• Increase to maximum 16 KB if large-block I/O size is predominant in the environment.
• With predominant small-block access, like 2 KB and 4 KB database environments, match cache page size to the predominant I/O size.

These are pages in write cache that have received new data from hosts but have not yet been flushed to disk. While a high value (i.e., a value between 60-80% of the write cache) for this measure is good as it increases the chance of a read coming from cache or additional writes to the same block of data being absorbed by the cache, a very high value – i.e., a value equal to or close to the total number of pages in the write cache – is a sign of bad health, as it indicates that the write cache is over-stressed.  

Ideally, the value of this measure should be high. A low value indicates that many read requests are serviced by direct disk accesses, which is a more expensive operation in terms of processing overheads.

If the read cache of the storage processor (SP) is enabled, then this measure will report the value Enabled. If not, then, this measure will report the value Disabled.

The numeric values that correspond to each of the states discussed above are available in the table below:

Note:

By default, this measure reports the above-mentioned Measure Values to indicate the status of the read cache. In the graph of this measure however, cache state is represented using the numeric equivalents - 0 or 1.

If the write cache of the storage processor (SP) is enabled, then this measure will report the value Enabled. If not, then, this measure will report the value Disabled.

The numeric values that correspond to each of the states discussed above are available in the table below:

Note:

By default, this measure reports the above-mentioned Measure Values to indicate the status of the write cache. In the graph of this measure however, cache state is represented using the numeric equivalents - 0 or 1.

For best performance, each Storage Processor (SP) should have the maximum amount of its memory in cache and should use the default settings for the cache properties. Therefore, ideally the number of memory pages in the cache should be high, as otherwise, storage system performance will suffer.

The read cache holds data that is expected to be accessed in the near future. If a request for data that is in the cache arrives, the request can be serviced from the cache faster than from the disks. Each request satisfied from cache eliminates the need for a disk access, reducing disk load. Typically, it would be good practice set the read cache to roughly 10% of available cache; 200 MB is the recommended minimum, and 1024 is the recommended maximum. For block-only VNX systems, the minimum can be set to 100 MB.

The initial read cache settings that EMC recommends for the different VNX models have been discussed in the table below:

If the workload exhibits a "locality of reference" behavior, where a relatively small set of data is accessed frequently and repeatedly, the read cache can improve performance. In read-intensive environments, where more than 70 percent of all requests are reads, the read cache should be large enough to accommodate the dataset that is most frequently accessed. For sequential reads from a LUN, data that is expected to be accessed by subsequent read requests is read (prefetched) into the cache before being requested. Therefore, for optimal performance, the read cache should be large enough to accommodate prefetched data for sequential reads from each LUN. An improperly sized read-cache can increase direct disk reads and can hence, adversely impact storage system performance.

If the read cache of the storage processor (SP) A is enabled, then this measure will report the value Enabled. If not, then, this measure will report the value Disabled.

The numeric values that correspond to each of the states discussed above are available in the table below:

Note:

By default, this measure reports the above-mentioned Measure Values to indicate the status of the read cache. In the graph of this measure however, cache state is represented using the numeric equivalents - 0 or 1.

Write cache serves as a temporary buffer where data is stored temporarily before it is written to the disks. Cache writes are far faster than disk writes. Also, write-cached data is consolidated into larger I/Os when possible, and written to the disks more efficiently. (This reduces the expensive small writes in case of RAID 5 LUNs.) Also, in cases where data is modified frequently, the data is overwritten in the cache and written to the disks only once for several updates in the cache. This reduces disk load. Consequently, the write cache absorbs write data during heavy load periods and writes them to the disks, in an optimal fashion, during light load periods. However, if the amount of write data during an I/O burst exceeds the write cache size, the cache fills. Subsequent requests must wait for cached data to be flushed and for cache pages to become available for writing new data. It is hence imperative that you rightly size the write cache and set cache watermarks appropriately. Cache watermarks control the flushing behavior of write cache. Given below are a few recommendations in this regard:

• Start with low watermark of 60% and a high watermark of 80%. This is suitable for a majority of the workloads.
• If frequent forced flushing occurs, reduce watermark values.
• Maintain a difference of about 20% between the low and high watermarks.
• Avoid drastic changes to these values unless advised by EMC Support.

If the write cache of the storage processor (SP) A is enabled, then this measure will report the value Enabled. If not, then, this measure will report the value Disabled.

The numeric values that correspond to each of the states discussed above are available in the table below:

Note:

By default, this measure reports the above-mentioned Measure Values to indicate the status of the write cache. In the graph of this measure however, cache state is represented using the numeric equivalents - 0 or 1.

For best performance, each Storage Processor (SP) should have the maximum amount of its memory in cache and should use the default settings for the cache properties. Therefore, ideally the number of memory pages in the cache should be high, as otherwise, storage system performance will suffer.

The read cache holds data that is expected to be accessed in the near future. If a request for data that is in the cache arrives, the request can be serviced from the cache faster than from the disks. Each request satisfied from cache eliminates the need for a disk access, reducing disk load. Typically, it would be good practice set the read cache to roughly 10% of available cache; 200 MB is the recommended minimum, and 1024 is the recommended maximum. For block-only VNX systems, the minimum can be set to 100 MB.

The initial read cache settings that EMC recommends for the different VNX models have been discussed in the table below:

If the workload exhibits a "locality of reference" behavior, where a relatively small set of data is accessed frequently and repeatedly, the read cache can improve performance. In read-intensive environments, where more than 70 percent of all requests are reads, the read cache should be large enough to accommodate the dataset that is most frequently accessed. For sequential reads from a LUN, data that is expected to be accessed by subsequent read requests is read (prefetched) into the cache before being requested. Therefore, for optimal performance, the read cache should be large enough to accommodate prefetched data for sequential reads from each LUN. An improperly sized read-cache can increase direct disk reads and can hence, adversely impact storage system performance.

Since the read cache is not mirrored, to use the available storage processor memory efficiently, ensure that you allocate the same amount of read cache to both the storage processors – i.e., A and B.

If the read cache of the storage processor (SP) B is enabled, then this measure will report the value Enabled. If not, then, this measure will report the value Disabled.

The numeric values that correspond to each of the states discussed above are available in the table below:

Note:

By default, this measure reports the above-mentioned Measure Values to indicate the status of the read cache. In the graph of this measure however, cache state is represented using the numeric equivalents - 0 or 1.

Write cache serves as a temporary buffer where data is stored temporarily before it is written to the disks. Cache writes are far faster than disk writes. Also, write-cached data is consolidated into larger I/Os when possible, and written to the disks more efficiently. (This reduces the expensive small writes in case of RAID 5 LUNs.) Also, in cases where data is modified frequently, the data is overwritten in the cache and written to the disks only once for several updates in the cache. This reduces disk load. Consequently, the write cache absorbs write data during heavy load periods and writes them to the disks, in an optimal fashion, during light load periods. However, if the amount of write data during an I/O burst exceeds the write cache size, the cache fills. Subsequent requests must wait for cached data to be flushed and for cache pages to become available for writing new data. It is hence imperative that you rightly size the write cache and set cache watermarks appropriately. Cache watermarks control the flushing behavior of write cache. Given below are a few recommendations in this regard:

• Start with low watermark of 60% and a high watermark of 80%. This is suitable for a majority of the workloads.
• If frequent forced flushing occurs, reduce watermark values.
• Maintain a difference of about 20% between the low and high watermarks.
• Avoid drastic changes to these values unless advised by EMC Support.

Since the write cache is mirrored, the write cache allocation applies to both the storage processors – i.e., A and B.

If the write cache of the storage processor (SP) B is enabled, then this measure will report the value Enabled. If not, then, this measure will report the value Disabled.

The numeric values that correspond to each of the states discussed above are available in the table below:

Note:

By default, this measure reports the above-mentioned Measure Values to indicate the status of the write cache. In the graph of this measure however, cache state is represented using the numeric equivalents - 0 or 1.

Each storage processor (SP) has a write cache in its memory, which mirrors the write cache on the other SP. Because these caches mirror each other, they are always either enabled or disabled, and always the same size. On powerup, a storage system automatically enables the write cache on each SP if the write cache size is non-zero.

Using this measure, you can determine whether the write cache of both SPs is currently enabled/disabled.

If the write cache is disabled, then this measure will report the value Enabled. If not, the measure will report the value Disabled.

The numeric values that correspond to each of the states discussed above are available in the table below:

Note:

By default, this measure reports the above-mentioned Measure Values to indicate the mirror status of the write cache. In the graph of this measure however, the mirror status is represented using the numeric equivalents - 0 or 1.