Cisco Cisco UCS B440 M1 High-Performance Blade Server Libro bianco
© 2012 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 10 of 106
●
Conversion from 3-node Oracle RAC to 12-node Oracle RAC
●
Implementation of Oracle 11.2.0.2 Automatic Storage Management (ASM) and Cluster Ready Services
(CRS) stack
●
Platform migration of 40 TB database with little downtime
●
Services enablement and failover strategies
●
Volume manager solution on Linux
●
Complete computing and Cisco UCS design
●
Implementation of networking switches and gateways with multiple paths
●
Implementation of storage infrastructure for production, nonproduction, and test systems
●
Backup, cloning, storage migration, refresh design, and optimization
C3 System Load Analysis
An analysis of the load for the C3 system is shown in Table 1.
Table 1.
C3 System Load Analysis
Category
Count
Load
Notes
Online applications
141
49%
●
All Java Virtual Machines (JVMs) connecting to C3
●
Service delivery applications load: 22.26%
●
Service sales applications load: 26.76%
Oracle Applications jobs
11
36%
Concurrent manager load
Others
49
9%
Not in any of the specified categories
User queries
143
4.5%
User-initiated queries
Administration
9
1%
Root, system, Oracle, Apache, WebSphere admin, apps install,
Oracle Cluster ready services administrator, Oracle Enterprise
Manager and other in-house tools administration
Oracle Cluster ready services administrator, Oracle Enterprise
Manager and other in-house tools administration
Business process integration and
management (BPIM) applications
management (BPIM) applications
38
< 1%
Load placed by TIBCO JVMs
Administrative jobs
11
< 1%
Replication jobs
Total
402
100%
Design Methodology
Metrics and Data Collected to Supplement the Design
●
Load and I/O metrics coming into C3 (for a 6-month period)
◦
Capture load and I/O statistics using the schema or OS user or execution module and align each
connection with an application.
◦
Slice the data based on technology and data affinity (type of data consumed by connection).
◦
Create a load metrics ratio for mapping the hardware (transactions per second [TPS]).
◦
Create an I/O metrics ratio for mapping the hardware I/O throughput.
◦
Based on the data affinity, technology, TPS, and I/O throughput, group applications for Oracle RAC
services.
◦
Align Oracle RAC services with hardware based on capacity benchmarks performed against hardware.
●
Historical data