Performance Impact Considerations - Blackberry ENTERPRISE SERVER FOR IBM LOTUS DOMINO - - PERFORMANCE BENCHMARKING GUIDE Manual

Performance impact considerations

When administering or implementing the BlackBerry Enterprise Solution on an IBM Lotus Domino server, perform
the following actions:
• Verify that the IBM Lotus Domino server can handle an increase in CPU, network, memory, and disk input and
output (I/O) activity.
• Verify that the IBM Lotus Domino server has enough disks to handle additional disk I/O activity to process all
message requests.
• Make sure that the disk subsystem can handle the additional disk transfers /sec.
Tip: To verify the disk subsystem I/O bandwidth, monitor the average disk queue length for direct attach
storage or disk seconds per read and seconds per write for SAN implementations. The average disk queue
length should not exceed one for each spindle and disk seconds per read, seconds per write should not exceed
50 ms on average, and spikes should not exceed 50 ms.
• Verify that no processor or I/O bottlenecks exist. RIM recommends keeping the operating system, Domino
data, Log.nsf, and transactions logs on separate disks.
• Monitor paging activity closely to identify any memory bottlenecks. Analyze pages /sec with available and
committed bytes counters. There must be at minimum 50 MB of available memory all the time.
• Before enabling BlackBerry user accounts, complete a baseline test of the system. To assess the impact of the
BlackBerry user accounts on the IBM Lotus Domino server infrastructure, compare the baseline performance
of the IBM Lotus Domino server to the performance of the IBM Lotus Domino server as you enable BlackBerry
user accounts.
The performance testing described in this document was performed under very specific load conditions and might
vary from site-to-site. RIM strongly recommends that you take baselines and monitor the relative impact of
enabling more BlackBerry user accounts or features.
6: IBM Lotus Domino performance test results
33