HP Command View EVA Software - Manual (Page 96)

initial copy data traffic for a replicated virtual disk is reported by the owning controller in the
source volume prefetch statistics.
HP EVA host connection
The HP EVA host connection object provides information for each HBA that has a connection to a
storage system.
The counters are:
•
Queue Depth , the average number of outstanding requests from each host adapter.
•
Busies , the number of host requests that are throttled by the storage system controller for
each host adapter when some internal storage system resource has reached a limitation.
HP EVA host port statistics
The EVA host port statistics object provides information about performance and data flow of
host-initiated activity on each storage system host port. Data replication traffic is not included in
these counters. Depending on the EVA storage system, there are either two ports per controller
(four per controller pair) or four host ports per controller (eight per controller pair).
The counters are:
•
Read Req/s —The number of read requests (per second) completed from each host port.
•
Read MB/s —The rate at which data is read from each host port.
•
Read Latency —The amount of time it takes to complete a read request (from initiation to
information receipt) through a host port. The time is an average of the read request latency
for all virtual disks accessed through this port, and includes cache hits and misses.
•
Write Req/s —The number of write requests (per second) completed from each host port.
•
Write MB/s —The rate at which data is written from each host port.
•
Write Latency —The amount of time it takes to complete a write request (from initiation to
information receipt) through a host port. The time is an average of the write request latency
for all virtual disks accessed through this port.
•
Average Queue Depth —The average number of outstanding host requests against all
virtual disks accessed through this host port. This number includes all host-initiated commands,
including non-data transfer commands.
HP EVA physical disk
The physical disk performance object reports information on each physical disk on the system that
is or has been a member of a disk group. These counters record all the activity to each disk and
include all disk traffic for a host data transfer, as well as internal system support activity. This activity
includes metadata updates, cache flushes, prefetch, sparing, leveling, snapclone and snapshot
support, and redundancy traffic. Metrics for each controller are reported separately. Therefore,
the total activity to each disk is the sum of activity to both controllers. The ID of each disk is from
an internal representation and is not definitive for disk identification.
The counters are:
•
Drive Queue Depth —The number of all active requests to each disk in the disk group,
over all the disks in the disk group.
•
Drive Latency —The amount of time from a data transfer command is sent to a disk to a
command completion notification is returned from the disk. The time is not separated into read
and write latencies. Completion of a disk command does not necessarily imply host request
completion, because the request to a specific physical disk can be part of a larger request
operation to a virtual disk. On the HSV100 series of controllers, average latency (the average
of read and write latencies) is reported. Starting with the HSV200 series of controllers, separate
metrics are provided for read and write latency.
96 Monitoring array performance using HP P6000 Performance Data Collector