HP XP P9000 Continuous Access Synchronous for Mainframe Systems User Guide

RecommendationDescriptionCondition
Spread the workload across several storage
systems to use additional channels.
The demand on the main system host channels
can affect performance.
High host channel
demand
Avoid performing restore operations to volumes
that belong to Cnt Ac-S Z pairs. Instead, restore
Cnt Ac-S Z operations can have a negative
impact on workloads with a high percentage of
Sequential write
operations
data to a scratch volume, and then create the
Cnt Ac-S Z pair.
sequential write operations, such as batch
processing operations (for example,
dump/restore, sort operations).
Consider increasing the cache size of the Cnt
Ac-S Z storage systems to improve overall
Large cache size improves read performance,
which allows more storage system resources to
Cache size
performance. For best results, the cache andbe devoted to write operations. Insufficient
NVS capacity of the main and remote systemcache resources result in command retries,
should be the same. This allows the remote site
to function adequately during disaster recovery.
state-change-pending (SCP) notifications, and
puncture conditions.
Distribute Cnt Ac-S Z remote copy operations
among several remote systems to avoid
overloading any one system.
The remote system’s performance directly affects
the performance of the main system. If a remote
system becomes overloaded with heavy update
Capacity of remote
system
activity, host and main system performance can
also be degraded.
Install a sufficient number of data paths to meet
all write-workload levels. This is especially
An inadequate number of data paths decreases
performance. Performing copy operations over
long distances also degrades performance.
Data paths
important for systems containing both P-VOLs
and S-VOLs.
Determining Round Trip Time
When you set up the Cnt Ac-S Z association between main and remote system, you specify a time
limit in milliseconds for data to travel from the P-VOL to the S-VOL. Round Trip Time is used by the
systems to control the initial copy pace when update copying is in progress. (To review the Round
Trip Time step, see “Configuring storage systems, defining logical paths” (page 58).)
This section provides instructions for determining your system’s Round Trip Time.
Note the following Round Trip Time considerations:
• If the difference between Round Trip Time and remote IO response time is significant, the
system slows or even interrupts the initial copy operation so that the update copy can continue.
Example of significant difference between the two: 1 ms RT Time : 500 ms remote IO response
time.
• If the difference between the two is insignificant, initial copying is allowed to run at the specified
pace.
Example of insignificant difference between the two: 1 ms : 5 ms
To determine Round Trip Time value
The value of Round Trip time= round trip time between MCU-RCU x 2(*) + initial copy response time (ms)
(*) A data transfer between main and remote system involves two response sequences in one command issuance.
Therefore, you need to double the round trip time.
• For the “round trip time” in the formula, please ask your service provider, or use a ping
command. If you do not use channel extenders between the main and remote system, specify
“1”.)
• The “initial copy response time” in the formula is the response time required for multiple initial
copy operations. With the following formula, determine the initial copy response time using
32 Planning for Continuous Access Synchronous Z