Consolidating Multiple Oracle Application Instances Considerations And Justifications
Sierra Atlantic, Inc. 34770 Campus Drive, Fremont, California 94555, USA Phone: 1.510-742-4100 Fax: 1.510-742-410 www.sierraatlantic.com
October 2002
1. Introduction Currently, there are many customers who are maintaining their complex and global business critical applications in latest release of Oracle Applications 11i. Across these organizations and installations it is understood that there may well be multiple servers, multiple databases and complex data dependencies. The final objective of these organizations is to get to a single server or small number of servers worldwide. They will want accounts for multiple countries stored within a single or small number of multiple organization enabled installations. Oracle Applications 11i provides an Internet platform and a consolidation approach to achieve this objective. Consolidation not only brings together a company’s IT processes and servers but also brings in changes that are much more fundamental. Companies can now truly step up to the challenge of globalization of hardware, software and business functions while still allowing security and autonomy to individual business areas.
2.
Consolidation Considerations
Hardware is becoming cheaper and more powerful for everyday computing, rapid increases in processor speed, backplane bandwidth, and disk capacity have made it possible for single enterprise servers to run a multitude of applications. The administrative support is expensive and hard to find. The applications, along with the servers needed to run them on, are more complicated than ever. Finding good support staff is difficult and retaining them harder still. By consolidating standard service support groups organizations can create a standardized and more efficient service to support each business unit in a consistent manner. Fundamentally Oracle has taken advantage of the Internet to provide many business functions, which can be tracked, entered and delivered globally. Vendor relationships become easier to handle, and organizations can also leverage a globally focused vendor more easily. Accurate cost models are easier to calculate in a centralized computing environment. 2.1 Management Considerations
Within the marketplace there are arguments for both the server distribution and server consolidation approaches. While management may see it as a quick way to reduce costs, users see it as a quick way to access all the data they need and neither group will allow any room for mistakes. It is the task of IT department to bring together all the groups involved and discuss and map out all the pros and cons. These may include the basic running costs of distributed vs. centralized computing but must also take into account political implications. Any organization centralizing their business functions will cause some internal friction, be it political or cultural. 2.1.1 Server Distribution considerations Business unit demand is rising for experimental use of applications and technologies in order to improve profits/revenues. Servers are evolving for single functions, such as e-mail, messaging, database or OLTP applications Distributed server life spans are falling as new servers supersede the old at an everincreasing rate The distributed server has a lower entry-level price point, and with usage of components like Intel-based servers, annual price performance improvements typically remain 40 percent
• • • •
Sierra Atlantic Inc
-1-
2.1.2
Server Consolidation considerations Cost savings through server consolidation has better economies of scale with more efficient use of equipment and tighter systems management. Efficient use of Systems and Administrative staff and overcome problems with IS staff shortages. Response times, end-users support, backup systems, and security, 24 X 7 availability will all receive improved service levels, if properly implemented. Vendor consolidation and use of the web in doing the business Consolidation of core functions such as receivables, data warehousing, ERP, and transaction will pave the way for efficient and accurate business data. Lower network and bandwidth costs leads to a viable solution for remote data processing. Cost Justifications
• • • • • •
2.1.3
Following are some of the cost justifications that should be considered:
• • •
• • •
2.2
Consolidation facilitates substantial reduction in staff strength such as operations and system support, helpdesk/network support, application development and LAN administrators. Constantly improving hardware hassled to an increase in processor, memory and storage utilization and the uptime of today’s computers, enabling large, centralized data centers, which are cheaper to maintain. Storage area networks (SAN) are reducing the amount of I/O configuration required and enabling disk farms rather than individual storage for each machine. Improvements in I/O channels etc within the servers are reducing the amount of CPU processing required to actually run the operating system and hardware. With fewer software vendors providing specialized enterprise applications the cost of software has fallen, with vendors offering greater discounts for larger orders. Systems will save on the cost of facilities with floor rental, power and security systems being combined into one area. Fewer software licenses are required.
Operational Considerations While there are many operational issues involved in server consolidation, the two most important factors that have maximum impact are high availability and backups. There are many things, which could cause a server to fail. These range from user errors such as dropping a table to natural disasters such as flooding. Other failures could occur in hardware or software. Whatever the disaster there have to be procedures in place to recover the lost data or loss of service. Providing a high availability server is neither simple nor inexpensive and the challenge is to understand what may go wrong, to examine the expected costs of downtime associated with these problems, and to compare those costs against the costs of various solutions to minimize downtime associated with these problems. 2.2.1 High availability options
Sierra Atlantic Inc
-2-
• • •
•
A single node (multiple CPU) in a single system image configured with mirrored raid disks and a dual fiber fabric can normally expect 99.5% availability during a year Many systems today come with various high-availability options, including redundant, reliable and hot swappable power supplies and fans and enhanced error correction. System management enhancements in many operating systems provide features aimed at reducing downtime. Some of these include configuration around failed CPU’s, preload and stage software upgrades, more on-line system tuning, single event log (for online error tracking) and a system console, which can provide offline error tracking. Clustering offers the highest level of availability by enabling configurations that eliminate single points of failure, i.e. the system itself is no longer a single point of failure. Clustering facilitates the least amount of unplanned downtime for applications, providing more that 99.95% availability. Backup options Disk mirroring is fast becoming the favored addition to tape backups. Straight disk nd rd mirroring can provide 2 and 3 copies of data that can be broken away and used to backup the system while leaving the ‘live’ system running. File system backups will inevitably overlap the database backups slightly, providing a second copy of the database configuration files and home directories. This improves Operations response to file restores while still maintaining the coherency of the Oracle instance backup
2.2.2
• •
3
Consolidation Business Justifications 3.1 One time expenses Initially there will be several one-time costs involved in consolidation. Consolidating applications will normally mean upgrading to the latest release of software and architecture for that application or enhancing the current release. Leases on both equipment and facilities may need to be terminated early in incurring fines. There will also be costs involved in negotiating new leases. New data center location will require additional equipment and communication lines. Apart from this, there will be one-time expenses to increase Bandwidth between sites (WAN) and one time consulting cost for merging Instances. 3.2 Savings due to Instance Consolidation Global Process Efficiencies Hardware savings Justifications 1. Hardware is becoming cheaper and more powerful. Improvements in processor, memory and storage peripherals have resulted in substantial reduction in Data center maintenance. Reduction in multiple Approx yearly cost savings Remarks
2.
3.
Sierra Atlantic Inc
-3-
Software Upgrade savings
Annual maintenance contracts (for each server). 4. Savings in data center floor space 1. One software upgrade instead of one upgrade for each instance
$10,000 to $250,000 for each instance upgrade
2. 3.
OS Upgrades, TAR management Software licenses
$10,000 savings for each instance
Savings due to support staff reduction
1.
2.
Savings due to Global Data warehouse solution Savings in thirdparty software interfaces support
1.
1.
Fewer staff is required to do Operations, System support, helpdesk and LAN administrators Alternately the staff can be re-assigned to other initiatives. Data warehouse system need not extract and merge data from multiple Instances. Interfaces will be integrated with one Instance only
$225,000 per annum per Instance
$10,000 for simple version upgrades (11.5.3 to 11.5.5) $250,000 for major technology stack upgrade like 10.7 to 11i It is assumed that each OS upgrade may result in an effort of 15 days including planning, execution etc. It is assumed that there will be a minimum three member staff reduction ($75,000/Annum) on operations front
At least 50% savings in Data Warehouse expenses and time At least 50% savings in Interfaces time and expenses.
Data merging can be avoided.
Savings due to Consolidated reporting, Business process Consolidation Savings due to consolidation of Inventory Organizations, Customers, Vendors and Employees Savings for New module implementations (HR, Payroll, Projects, iStore, Advance pricing etc)
1. 2. 3. 1.
Real time consolidated view of Financial information Standardize financial delegations/Methods Standardize business process Facilitates deployment of Ebusiness initiatives like iProcurement, iStore and other eCommerce initiates
Substantial savings due availability of real time Financial data across Geographies Substantial savings in implementations
There is no need to duplicate integration with multiple instances and savings on thirdparty software licenses There is no need to extract and consolidate data to get Global view. Uniform notifications, approvals and workflows
1.
Global Instances require a single implementation instead of multiple implementations
$75,000 for each new module per Instance
It is assumed that each new module implementation will cost around $75,000 per instance.
Sierra Atlantic Inc
-4-
4
Summary
In essence, it is understood that the largest potential ROI (Return on Investment) is possible when consolidating servers. As explained, there are a number of one-time and ongoing additional costs, some of which may be hard to anticipate or identify. Standard reporting against a single installed server provides automatic consolidation of financial data within the reports while still providing the flexibility to provide each separate organization with it’s own subset of reporting information. Multiple installs will provide individual organizations with reports, however the consolidation of data for the overall organization will be complex and require manual intervention. There are considerable advantages for having a single server running Oracle Applications across all Business Units. This reduces the cost of running and maintaining data centers in multiple locations and makes global reporting relatively straightforward.
Sierra Atlantic Inc
-5-