HP MPX200 Multifunction Router - Manual (Page 139)

7 Performance and best practices
This chapter discusses the factors affecting data migration solution performance and offers
suggestions for obtaining maximum performance.
Performance factors
DMS provides maximum throughput of 4 TB per hour. The migration performance depends upon
several factors, including:
•
Number of concurrent migration jobs running on the MPX200
•
I/O size used for data transfer
•
I/O traffic serviced by the array for other applications active during data migration
•
FC link speed
•
RAID configuration of LUNs
•
Amount of host I/O on the LUN
Maximizing performance
Suggestions for maximizing performance include the following:
•
Use Serial Scheduling of migration jobs, see “Using the data migration wizard” (page 55)
instead of simultaneously starting all jobs.
•
The MPX200 is capable of simultaneously configuring a maximum of 256 jobs on each blade.
However, HP recommends that you simultaneously run only four to six jobs. The remaining
jobs can be scheduled using Serial Scheduling or Delayed Scheduling.
•
LUNs belonging to different RAID groups (sets of physical drives) or on a different array
controller on the storage array should be scheduled to run simultaneously. To improve write
performance, disable write cache settings on the destination array.
•
To complete data migration jobs faster, run the migration job during off-peak load hours. For
online jobs, the source and destination LUNs are synchronized until the migration job is
acknowledged.
Optimal configuration and zoning
To get the best performance from the router, configure your system as follows:
•
In the physical topology, configure switches with two ports each from the router, source array,
and destination array.
•
Set the zoning such that each router port sees both controller ports from the source array and
the destination array.
•
Balance the LUNs on the source and destination arrays between the two controller ports.
NOTE: HP recommends that you do not simultaneously run more than four to six data migration
jobs per blade.
Expected time of completion (ETC) for data migration jobs
Overview
The MPX200 can help determine expected time of completion of a job currently in the Run State
or the Completed state if it is configured for online migration. This feature is applicable to all types
of jobs: online and offline, local and remote, as well as LUN compare jobs. The value is displayed
Performance factors 139