128 Storage Management with DB2 for OS/390
Cache Subsystem Device Overview
This report lists the devices known by the subsystem at the beginning of the
interval. Each line displays statistics for a specific (functional) volume. The I/O
rate, divided into twogroups (CACHE HIT and DASD I/O), shows the different
types of I/O activity in each group. The *ALL line consolidates values at the
subsystem level.The fields to review are, in decreasing order of importance:
• I/O RATE, number of I/O requests per second.
• % READ, percentage of read requests compared to all read plus write
requests. When combined with DEVICE ACTIVITYRATE of LCU level
DEVICE report, this value enables you get to the WRITE ACTIVITYRATE by:
• (100 - %READ) * DEVICE ACTIVITY RATE / 100
• The write activity rate value is an impor tant factor forper formanceevaluation
in a remote copyenvironment.
• READ H/R, read hit ratio.
• WRITE H/R, write hit ratio.
• DFW, rate of DFW requests.
• STAGE, rate of any (normal or sequential and read or write) I/O requests with
cache miss.
• ASYNC RATE, number of tracks asynchronously destaged from cache to disk
as a naturalc onsequenceof least recently used cache management
algorithms.
• ICL, rate of inhibit cache load requests should be at zero. See 9.3.6, “No
Caching—Inhibit Cache Load” on page 92.
11.2.1.2 Direct Access Device Activity Report
This report can be produced at either the LCU level (standard) or the Storage
Group levelfor each LPAR. Both reporting levels should be used. Although the
LCU report is relevant with CACHErepor ts,Storage Group level reporting
automatically consolidates information in a global view of I/O activity consistent
with the installation organization defined to SMS. With appropriate SMS
definitions, such StorageGroup level reporting should map the applications point
of view.
Toget standard LCU reporting, specify:
REPORTS( DE VICE( DASD ) )
Toget Storage Group reporting, specifyin the RMF Monitor III postprocessor:
REPORTS ( DEVICE ( SG ( storage-group-name) ) )
Figure 50 on page 129 shows the Direct Access Device Activity Report at the
Storage Group (SG) level.
The response time of a specific volume in a given LPARconsists of service time
and volume thread queuing time. Service time splits into:
•Pendingtime, which covers the channel subsystem and connections delays to
the LCU due to ESCON Director switching, and, in ESCON multiple image
facility (EMIF), channel sharing between LPARs