JVM Options:
 -server -Xms3300m -Xmx3300m -Xmn1024m -Xss128k -XX:+AggressiveOpts
 -XX:+UseParallelGC -XX:ParallelGCThreads=32 -XX:PermSize=128m
 -XX:LargePageSizeInBytes=4m -XX:+UseParallelOldGC
 -verbose:gc -XX:+PrintGCDetails -XX:+PrintGCTimeStamp
 -XX:-UseAdaptiveSizePolicy -XX:+PrintAdaptiveSizePolicy
 -XX:MaxTenuringThreshold=15 -XX:InitialSurvivorRatio=10
 -XX:SurvivorRatio=10 -XX:TargetSurvivorRatio=90
 -XX:ReservedCodeCacheSize=64m -XX:CICompilerCount=4
 -Dweblogic.management.discover=false
 -Dweblogic.SocketReaders=4 -Dweblogic.StuckThreadMaxTime=900
 -Dweblogic.diagnostics.debug.DebugLogger.DISABLED=true
 -Doracle.jdbc.defaultRowPrefetch=300
Java process started in FX class using /usr/bin/priocntl -e -c FX
Java process started with LD_PRELOAD_32=/usr/lib/libumem.so
Oracle patch for WebLogic Server Standard Edition 10.3.3: 9686209

JVM Options:
 -d64 -server -Xms20g -Xmx20g -Xmn3g -Xss256k -XX:+AggressiveOpts
 -XX:+UseParallelGC -XX:ParallelGCThreads=32 -XX:LargePageSizeInBytes=4m
 -verbose:gc -XX:+PrintGCDetails -XX:+PrintGCTimeStamps
 -XX:-TraceClassUnloading  -XX:PermSize=128m -XX:MaxPermSize=512M
 -XX:+UseCompressedOops -XX:TargetSurvivorRatio=90 -Dweblogic.SocketReaders=4
 -Dsun.net.inetaddr.ttl=1 -Dnetworkaddress.cache.ttl=1
Java process started in FX class using /usr/bin/priocntl -e -c FX
Java process started with LD_PRELOAD_64=/usr/lib/sparcv9/libumem.so

 db_name                         = spec
 control_files                   = ?/dbs/cntrlspec
 compatible                      = 11.1.0.6.0
 aq_tm_processes                 = 0
 cpu_count                       = 384
 db_4k_cache_size                = 48G
 db_8k_cache_size                = 30G
 db_block_checksum               = FALSE
 db_block_checking               = FALSE
 db_block_size                   = 2048
 db_cache_advice                 = OFF
 db_cache_size                   = 202207M
 db_file_multiblock_read_count   = 128
 db_files                        = 256
 db_keep_cache_size              = 5G
 db_recycle_cache_size           = 60G
 db_writer_processes             = 32
 disk_asynch_io                  = TRUE
 dml_locks                       = 1000
 filesystemio_options            = setall
 log_buffer                      = 201326592
 log_checkpoint_interval         = 0
 log_checkpoints_to_alert        = TRUE
 open_cursors                    = 2000
 parallel_max_servers            = 256
 processes                       = 12000
 query_rewrite_enabled           = FALSE
 replication_dependency_tracking = FALSE
 resource_manager_plan           = ''
 sessions                        = 25000
 shared_pool_size                = 24G
 statistics_level                = BASIC
 timed_statistics                = FALSE
 trace_enabled                   = FALSE
 transactions                    = 75000
 transactions_per_rollback_segment = 1
 undo_management                 = AUTO
 undo_retention                  = 300
 undo_tablespace                 = undo_ts

Oracle processes run in 16 processor sets using psrset -b pset PID.
and each process run in the FX class using /usr/bin/priocntl -s -c FX -i pid PID.
Oracle Log Writer run in a processor set containing 1 core.
using psrset -b pset pid and run in the RT class using
/usr/bin/priocntl -s -c RT -p 1 -i pid.
A separate processor set with 21 processors configured to process
interrupts from 20 network cards and the clock interrupt.
A processor set with 2 processors used to process interrupts
for the disk controllers configured for the redo logs.
The processor set configuration is in the FDA.
Oracle patches for 11.1.0.7: p7627743, p8570516.

JVM Options:
 -server -XX:+AggressiveHeap -XX:+UseParallelGC
 -XX:LargePageSizeInBytes=4m -XX:-TraceClassUnloading
 -Xms2g -Xmx2g -Xmn600m -XX:+DisableExplicitGC -Xbatch
 -Xss128k -Dsun.net.inetaddr.ttl=0 -Dnetworkaddress.cache.ttl=0
 -Dweblogic.ThreadPoolSize=64 -Dweblogic.ThreadPoolPercentSocketReaders=80

System Tuning in /etc/system:
set ip:ip_soft_rings_cnt = 16
set ip:ip_squeue_fanout = 1

Network Tuning:
ndd -set /dev/tcp tcp_conn_req_max_q 40000
ndd -set /dev/tcp tcp_conn_req_max_q0 40000
ndd -set /dev/tcp tcp_time_wait_interval 10000
ndd -set /dev/tcp tcp_xmit_hiwat 524288
ndd -set /dev/tcp tcp_recv_hiwat 524288
ndd -set /dev/tcp tcp_smallest_anon_port 4096
ndd -set /dev/tcp tcp_naglim_def 1
ndd -set /dev/tcp tcp_fin_wait_2_flush_interval 10000

Tuning in /kernel/drv/ixgbe.conf:
tx_queue_number=12;
intr_throttling=2000;

Disk configuration:
2x 73GB SAS disks for OS and application software using UFS
SVM mirror using 2x SSD disks for logs and JMS persistence filestore.
Configuration notes:
Eight application server instances were hosted in 4 separate Solaris Containers.
The Four Containers were bound to separate processor sets where each pset
consisted of 56 HW threads (7 cores) using /usr/sbin/psrset -b PSET PID.
Interrupts were enabled on remaining HW strands in the default processor set.
The Solaris Container and Processor Set configuration is in the FDA.
Each Container on the system was connected to 2 subnets.
Four Containers shared one of two 10Gbit NICs.
Two Containers shared one of four 1 Gbit NICs.
The first subnet in each container was used for driver access using the 10 Gbit.
The second subnet in each container was used for database access
using the 1GBit NIC.

System Tuning in /etc/system:
set kernel_cage_enable=0
set autoup 345600
set hires_tick = 1
set ip:ip_squeue_fanout = 1
set ip:ip_soft_rings_cnt = 8
set mmu_ism_pagesize = 0x10000000
set lgrp_mem_pset_aware=1
set lgrp_mem_default_policy=3
Disabled hardware prefetch by 'setprefetchmode -s off' command on XSCF

Interrupts from the Network and Disk controllers were configured
using the /usr/sbin/pcitool utility
Network Tuning:
ndd -set /dev/tcp tcp_conn_req_max_q 40000
ndd -set /dev/tcp tcp_conn_req_max_q0 40000
ndd -set /dev/tcp tcp_time_wait_interval 10000
ndd -set /dev/tcp tcp_xmit_hiwat 524288
ndd -set /dev/tcp tcp_recv_hiwat 524288
ndd -set /dev/tcp tcp_smallest_anon_port 4096
ndd -set /dev/tcp tcp_naglim_def 1
ndd -set /dev/tcp tcp_fin_wait_2_flush_interval 10000

Shared memory settings in /etc/project:
user.oracle:101:Oracle DBA:::project.max-shm-memory=(privileged,549755813888,deny)

Network configuration:
20x 1GBit NIC
1x 10GBit NIC

External Disk configuration:
6 Sun Storage F5100 microSSDArray (80 x 24GB Array)
24x DualPort SAS HBA
Solaris patch 141874-05

System Tuning in /etc/system:
set ip:ip_soft_rings_cnt = 16

Network Tuning:
ndd -set /dev/tcp tcp_conn_req_max_q 40000
ndd -set /dev/tcp tcp_conn_req_max_q0 40000
ndd -set /dev/tcp tcp_xmit_hiwat 524288
ndd -set /dev/tcp tcp_recv_hiwat 524288
ndd -set /dev/tcp tcp_naglim_def 1
ndd -set /dev/tcp tcp_time_wait_interval 10000
ndd -set /dev/tcp tcp_fin_wait_2_flush_interval 10000
ndd -set /dev/tcp tcp_smallest_anon_port 4096

Blade Chassis Configuration:
1x Sun Blade 6000 Chassis each configured with
10x Sun Blade X6270 Server Modules and
10x Dual Port 10Gbit PCIe Express Modules

System Tuning in /etc/system:
set rlim_fd_cur=262144

Network Tuning:
ndd -set /dev/tcp tcp_conn_req_max_q 40000
ndd -set /dev/tcp tcp_conn_req_max_q0 40000
ndd -set /dev/tcp tcp_xmit_hiwat 524288
ndd -set /dev/tcp tcp_recv_hiwat 524288
ndd -set /dev/tcp tcp_naglim_def 1
ndd -set /dev/tcp tcp_time_wait_interval 10000
ndd -set /dev/tcp tcp_fin_wait_2_flush_interval 10000
ndd -set /dev/tcp tcp_smallest_anon_port 4096

Network Tuning:
ndd -set /dev/tcp tcp_conn_req_max_q 16384
ndd -set /dev/tcp tcp_conn_req_max_q0 16384
ndd -set /dev/tcp tcp_xmit_hiwat 131072
ndd -set /dev/tcp tcp_recv_hiwat 131072
ndd -set /dev/tcp tcp_naglim_def 1
The DNS named process run in the RT class using /usr/bin/priocntl -s -c RT -p 1 -i pid PID