Determining SuperCluster M6-32 Configurations
Determine the Number of Compute Servers
Determine the Number of DCUs in Each Compute Server
Determine the Number of CMUs in Each DCU
Determine the Amount of Memory in Each DCU
Determine the PDomain Configuration on Each Compute Server
Determine the LDom Configuration for Each PDomain
Determining the Best Configuration for Your Situation
Understanding PDomain Configurations
Allocating CPU Resources for LDoms
Allocating Memory Resources for LDoms
Understanding PCIe Cards and Slots for LDoms
Understanding Storage for LDoms
Understanding SuperCluster M6-32
Identifying SuperCluster M6-32 Components
Understanding DCU Configurations
Understanding Half-Populated DCU Root Complexes
Understanding Fully-Populated DCU Root Complexes
Extended Configuration PDomain Overview
Understanding Extended Configuration PDomains
Understanding Base Configuration PDomains
Understanding Compute Server Hardware and Networks
Understanding LDom Configurations for Extended Configuration PDomains
Understanding LDom Configurations for Base Configuration PDomains
Understanding Clustering Software
Cluster Software for the Database Domain
Cluster Software for the Oracle Solaris Application Domains
Understanding System Administration Resources
Understanding Platform-Specific Oracle ILOM Features
Oracle ILOM Remote Console Plus Overview
Oracle Hardware Management Pack Overview
Time Synchronization and NTP Service
Multidomain Extensions to Oracle ILOM MIBs
Hardware Installation Overview
Hardware Installation Task Overview
Hardware Installation Documents
Preparing the Site (Storage Rack and Expansion Racks)
Prepare the Site for the Racks
Network Infrastructure Requirements
Compute Server Default Host Names and IP Addresses
Compute Server Network Components
Storage Rack Network Components
Cable the ZFS Storage Appliance
ZFS Appliance Power Cord Connection Reference
ZFS Storage Appliance Cabling Reference
Leaf Switch 1 Cabling Reference
Leaf Switch 2 Cabling Reference
IB Switch-to-Switch Cabling Reference
Cable the Ethernet Management Switch
Ethernet Management Switch Cabling Reference
Connect SuperCluster M6-32 to the Facility Networks
Expansion Rack Default IP Addresses
Understanding Internal Cabling (Expansion Rack)
Understanding SuperCluster Software
Identify the Version of SuperCluster Software
Controlling SuperCluster M6-32
Powering Off SuperCluster M6-32 Gracefully
Power Off SuperCluster M6-32 in an Emergency
Monitoring SuperCluster M6-32 (OCM)
Monitoring the System With ASR
Configure ASR on the Compute Servers (Oracle ILOM)
Configure SNMP Trap Destinations for Storage Servers
Configure ASR on the ZFS Storage Appliance
Configuring ASR on the Compute Servers (Oracle Solaris 11)
Approve and Verify ASR Asset Activation
Change ssctuner Properties and Disable Features
Configuring CPU and Memory Resources (osc-setcoremem)
Minimum and Maximum Resources (Dedicated Domains)
Supported Domain Configurations
Display the Current Domain Configuration (osc-setcoremem)
Display the Current Domain Configuration (ldm)
Change CPU/Memory Allocations (Socket Granularity)
Change CPU/Memory Allocations (Core Granularity)
Access osc-setcoremem Log Files
Revert to a Previous CPU/Memory Configuration
Remove a CPU/Memory Configuration
Obtaining the EM Exadata Plug-in
Known Issues With the EM Exadata Plug-in
Configuring the Exalogic Software
Prepare to Configure the Exalogic Software
Enable Domain-Level Enhancements
Enable Cluster-Level Session Replication Enhancements
Configuring Grid Link Data Source for Dept1_Cluster1
Configuring SDP-Enabled JDBC Drivers for Dept1_Cluster1
Create an SDP Listener on the IB Network
Administering Oracle Solaris 11 Boot Environments
Advantages to Maintaining Multiple Boot Environments
Mount to a Different Build Environment
Reboot to the Original Boot Environment
Create a Snapshot of a Boot Environment
Remove Unwanted Boot Environments
Monitor Write-through Caching Mode
There are two main approaches to modifying resource allocations:
All resources allocated – You move resources from domains to other domains, and ensure that all resources are allocated.
Some resources are unallocated – You allocate less than the maximum available cores and memory for a compute node. Any unused cores are considered parked cores and are not counted for licensing purposes. However, parked cores are added to the logical CPU and memory repository. If you have Root Domains, you can later allocate the repository resources to I/O Domains. See Park Cores and Memory.
Depending on which command you use to view domain resources, you might need to convert socket, core, and VCPU values.
|
See one of these procedures:
In this example, one compute node on a SuperCluster M6-32 has two dedicated domains and two Root Domains.
|
Calculating the total amount of CPU and memory resources gives you a starting point for determining your resource plan.
While identifying resources, keep these points in mind:
Root Domain resources – Are a small amount of resources that are reserved for the exclusive use of Root Domains. Do not factor these resources into your plan.
Unallocated resources – These resources are placed in the logical CPU and memory repositories when Root Domains are created, or by leaving some resources unallocated when you use the osc-setcoremem command.
In this example, the resources for the dedicated domains and the unallocated resources are summed to provide total resources. The Root Domain resources are not included in total resources.
|
In this example, 12 cores and 1 TB memory are parked from the primary domain, and 18 cores and 1536 GB memory are parked from the ssccn3-dom1 domain.
The total resources for before and after columns should match. This check ensures that all resources are accounted for in your plan.
|
Change resource allocations at the socket granularity level.
Change resource allocations at the core granularity level.
Increase unallocated resources.