The mainframe has been around for decades and is considered by many to be the ultimate legacy platform for
scale-up computing. These monolithic systems have proven their ability to deliver the performance, scalability,
security, manageability, and high availability large businesses demand for their core transactional and database
applications. Yet the high costs associated with scaling and adapting mainframe environments have many companies
looking for ways to extend or replace their mainframes with alternative platforms that improve agility and reduce
costs without sacrificing quality of service (QoS).
Since moving workloads off a mainframe often represents a fundamental shift in IT strategy, a comprehensive and systematic approach is needed to minimize cost and risk. This paper describes a methodology (Figure 1) that has been used successfully for mainframe migrations. It offers a useful starting point for business and IT decision makers looking for ways to improve IT capability and agility, without sacrificing QoS and without incurring the high cost of expanding an existing mainframe environment.
Re-Hosting Mainframe Applications on
Intel® Xeon® Processor-based Servers
A Proven Methodology for Mainframe Migration
1
2
3
4 Workshop
Assessment
Migration
Methodology
Approach Solutions
Overview Business &
Financial Overview
Evaluate &
Analyze Plan
Definition Solution
Design
Proof of
Concept Solution
Test System
Optimization
Infrastructure
Deployment Application
Roll-Out Operations &
Monitoring Deployment
Figure 1. Following a proven migration methodology helps to reduce cost and risk and accelerate time to benefits.
Getting Started
To ensure business and technical issues are thoroughly explored, an interdisciplinary approach is necessary. The initial
project team should include business managers, business analysts, system and application architects, and IT operational staff. Support from executives is also imperative because a mainframe migration will impact multiple business units.
Unless the team has experience with mainframe migrations, engaging with a consultant is highly recommended. Many system integrators, independent software vendors (ISVs), and hardware suppliers have proven sets of tools and extensive experience with mainframe migrations.
A technical and planning workshop from suppliers should be a first step. It will provide a solid foundation for an efficient and effective migration. Outsourcing some or all of the migration may also be a good choice, depending on time constraints and the availability of internal resources.
Establishing Goals
Establishing clear goals is important to keep the project on track and aligned with business priorities and needs. Goals should be specific and, when possible, quantitative to provide a framework for weighing alternatives throughout the planning process. Issues to consider include business growth, cost reduction, IT efficiency and agility, and governance, including risk management. A mainframe migration of modern applications, such as Java-based solutions, can deliver significant value across all these areas.
Understanding business and IT priorities is critical to ensure that projects remain aligned with the needs of the business.
Assessing the Current Environment Once goals are established, a comprehensive assessment of the current business, technical, and operational infrastructure environments must be performed (Figure 2). This assessment should be comprehensive and thorough because it provides the foundation for understanding end-to-end solution requirements. It should include physical, logical, and
solution architectures across all systems.
It should also include production, staging, disaster recovery, quality assurance, and development environments, as well as key mainframe interfaces, system interdependencies, and performance bottlenecks. Future requirements should also be evaluated with respect to the current solution.
The IT industry has developed a number of infrastructure assessment and
inventory tools to help with hardware and application migrations. These tools range from simple spreadsheets to advanced programs that may include automated data center discovery. Depending on the selected tool, it may measure system workloads, resource utilization, and dependency mapping. It may also discover physical and virtual IT assets, applications, and the relationships between them. The resulting assessment typically includes detailed reports and graphs, and can be useful for system design and for identifying opportunities to virtualize infrastructure and reduce overall costs.
Figure 2. A comprehensive and integrated assessment of the existing infrastructure provides a foundation for
Integrated Assessment
Business Solutions & Applications
Staging & Development Systems Backup Systems Storage Devices Network I/O
Database Application Services Commercial and Custom Applications
Operating System
Enterprise Infrastructure
Software Utilities, IT Operational Tools IT Operations & Management IT Management Framework
Hardware Devices, Software Drivers Infrastructure Performance
System Performance
Integrated Solution and Infrastructure Assessment
Integrated System Assessment
Mainframe Migration in the Real World
HSBC
One of the World’s Largest Banks When heavy workloads from a new suite of loan applications created performance and reliability issues for an HSBC mainframe in Mexico City, the company had a choice to make: either upgrade the mainframe or migrate existing workloads onto an alternative platform. After a thorough review of alternatives and consultation with Intel, HSBC Mexico used the methodology outlined in this paper to re-host the applications on Intel® Xeon® processor- based servers running Linux*. The resulting migration met or exceeded all of HSBC’s goals.
• Performance and uptime were improved, delivering a better experience to customers and branch office personnel.
• Mainframe workloads were reduced by 2,000 MIPS and monthly service charges by 70 percent.1,†
• Business flexibility was increased because the new solutions made it easier to scale and adapt applications.
For a detailed technical discussion of the HSBC mainframe migration, read the Intel white paper at www.intel.com/content/www/us/
en/mission-critical/mission-critical- meeting-todays-it-challenges.html
Figure 3. Web server performance has increased substantially for two-socket Intel® Xeon®
processor-based servers. The latest systems based on the Intel® Xeon® processor 5600 series are ideal for processing large numbers of concurrent user requests, as demonstrated by published results on the SPECWeb*2005 benchmark.2,†
60,000
Intel® Xeon processor 3.80 GHz (2 MB L2, 800 MHz FSB, single core)
4555 15193
29591
104422 83198
30,000
0 120,000
SPECweb*2005 Score
(Higher is better)
90,000
Intel® Xeon processor 5160 (3.0 GHz, 4 MB L2, 1333 MHz FSB, dual core) Intel® Xeon processor X5460 (3.16 GHz, 2x6 MB L2, 1333 MHz FSB, quad core) Intel® Xeon processor X5570 (2.93 GHz, 8 MB L3, 6.4 GT/s, quad core) Intel® Xeon processor X5680 (3.33 GHz, 12 MB L3 cache, 6.4 GT/s, six core)
Creating the Proposal
Creating a proposal will typically require evaluating multiple options for meeting current and future needs. Should custom applications be rewritten, re-factored, or replaced? Should new infrastructure be scaled up, scaled out, or virtualized?
Should a private cloud architecture be considered? Technical feasibility should be evaluated for each option, including the challenges associated with transitioning to the new solution. If some applications will remain on the mainframe, it will be necessary to explore options for integration, including hardware, software, and operational issues.
Once technical feasibility is confirmed, return on investment (ROI) and total cost of ownership (TCO) analyses should be performed for the solutions that are most feasible and have the highest business migration priorities. This will help quantify the decision-making process. Technical and financial assessments should include sufficient detail to identify significant risks that could potentially impact the success of the project.
Developing the Solution Architecture An interdisciplinary approach is required to develop a detailed solution architecture and migration plan. The team should include system architects, application architects, data center IT operations personnel, database administrators, and quality assurance staff. It should also include business unit representatives to ensure the proposed architecture is well aligned with current and projected business needs.
IBM and other mainframe vendors offer detailed and proven design options for deploying both horizontally and vertically scaled solutions. In many cases, a two- tier or three-tier architecture will be optimal. Selecting best-fit Intel® Xeon®
processor-based servers for the new solution will depend on specific workloads, performance requirements, and solution architectures. The following guidelines will be helpful in many scenarios.
• Web Servers. Web requests can be distributed efficiently across multiple two-socket Intel® Xeon® processor 5600 series-based servers. Performance has increased substantially in the two-socket server space in recent years, and today’s systems are well suited for processing large numbers of concurrent user requests (Figure 3).†
• Application and Database Servers.
Four-socket, eight-socket, and larger servers based on the Intel® Xeon®
processor E7 family provide scalable performance per system to support heavy application and database workloads. With up to two terabytes of memory and a high-speed, point-to- point interconnect system, these servers support exceptionally fast processing for high-volume transaction workloads (Figure 4).† The tightly-integrated interconnect system also provides reliability, availability, and serviceability (RAS) for mission-critical environments through features such as advanced error correction and automated link recovery.
Unless the mainframe is being decommissioned, it may be necessary to access mainframe data and manage transactions across the new systems and solutions and the mainframe environment. Many options are available, such as IBM CICS* Transaction Gateway and IBM MQ* Series message-oriented middleware. These and other options are easily deployed on Intel Xeon processor-based servers. They provide flexible, scalable, and secure options for accessing information and managing transactions and applications on distributed systems.
Figure 4. Four-socket, eight-socket, and larger Intel® Xeon® processor-based servers deliver scalable performance for demanding Java applications, as demonstrated by published results on the SPECjAppServer2004 and SPECjbb2005 benchmarks.†
16,000
Intel® Xeon processor X7350 (8 M cache, 2.9 GHz, 1066 MHz FSB) 3340 4410
11057
20092
8,000
0
SPECjAppServer*2004 JOPS@Standard
(Higher is better) 24,000
Intel® Xeon processor X7460 (16 M cache, 2.66 GHz, 1066 MHz FSB Intel® Xeon processor X7560
(24 M cache, 2.26 GHz, 6.40 GT/s Intel® QPI) Intel® Xeon processor X7560
(24 M cache, 2.26 GHz, 6.40 GT/s Intel® QPI)
Intel® Xeon processor X714OM (16 M cache, 3.40 GHz, 800 MHz FSB) Intel® Xeon processor X7350 (8 M cache, 2.93 GHz, 1066 MHz FSB) Intel® Xeon processor X7460 (16 M cache, 2.66 GHz, 1066 MHz FSB) Intel® Xeon processor X7560
(24 M cache, 2.26 GHz, 6.40 GT/s Intel® QPI) Intel® Xeon processor X7560
(24 M cache, 2.26 GHz, 6.40 GT/s Intel® QPI) Intel® Xeon processor X7560
(24 M cache, 2.26 GHz, 6.40 GT/s Intel® QPI) 217334 537116 633897
8-Socket 4-Socket
8-Socket 4-Socket
3816799
2021535
SPECjbb*2005 BOPS
(Higher is better) 4,000,000
3,500,000 3,000,000 2,500,000 2,000,000 1,500,000 1,000,000 500,000 0
Intel® QPI - Intel® QuickPath Interconnect
based on the Intel Xeon processor E7 family deliver high-end scalability and advanced reliability for both vertically and horizontally scalable solutions, and are well suited for a broad range of workloads currently running in many mainframe environments.
A successful migration requires detailed planning, commitment, a team effort by multiple business and technical teams, and direct support from senior management.
Applying a comprehensive, step-by- step approach using a proven migration methodology is essential for coordinating business and IT efforts to deliver desired benefits, while managing costs and risks.
Professional services and support are available from leading system integrators, ISVs, and hardware suppliers around the world, many of whom have extensive experience in mainframe migration.
Proof-of-Concept (PoC) Testing When migrating custom applications, PoC testing will help technical teams identify and address potential design and operational issues early in the design and migration processes. For example, targeted PoCs may be appropriate to validate performance, scalability, transaction management, and data access for the new solution. However, extensive PoCs may not be necessary when migrating commercial applications.
ISVs can often provide reference designs and sizing guides, as well as best practice recommendations that have been proven worldwide across numerous customer implementations.
Moving to Production
In moving to production, a staged and phased transition is generally preferable.
A pilot deployment using a subset of applications can allow IT to identify and address problems before they affect the
broader business environment. It can also provide IT with a good opportunity to integrate the new solution into the broader development and quality assurance environments and to integrate and test new operational tools, scripts, utilities, and processes. A successful pilot may require considerable collaboration among multiple groups and teams, and time and dedicated resources should be allocated accordingly.
Conclusion
Migrating modern workloads from a mainframe to Intel Xeon processor- based servers can deliver fundamental advantages, including better performance and scalability, improved business and IT agility, and lower capital and operating costs. Businesses have been migrating mainframe applications onto distributed, horizontally scalable solutions for years.
Newer eight-socket and larger servers
ASSESSING THE CURRENT ENVIRONMENT
Perform a comprehensive evaluation and documentation of:
Backup for one Windows Small Busi Physical, logical, and solution architectures
Performance and functionality across operational, development, staging, disaster recovery, and quality assurance infrastructure System interfaces, interdependencies, networking, and storage
Software solution compatibility, business applications and solutions, OS, middleware, Java* Virtual Machine (JVM), database, and database connections
Application, database, and transaction performance, including current bottlenecks
Future performance requirements with projected growth, workloads, transactions, and new services Security, compliance, and IT management and operational framework
DESIGNING THE NEW SOLUTION AND SYSTEM ARCHITECTURE
Develop the solution architecture, including physical and logical systems that account for:
Software stack, including business applications and solutions, Enterprise Service Bus (ESB), application server and JVM, database systems, and OS Compute, network, and storage systems, including topology for load balancing and for horizontal and vertical scaling
Performance and scalability for current and projected workloads System availability and maintainability
Security, compliance, and integration with IT management framework
Establish solutions for connecting to the mainframe environment, including:
Java database connectivity (JDBC) between the application server (IBM WebSphere* Application Server, JBOSS, Oracle WebLogic* Server, and so on) and mainframe databases
Transaction management (CICS Client Gateway, and so on)
Message passing connectivity (IBM WebSphere MQ Client, IBM WebSphere MQ Server, and so on) PLANNING FOR MIGRATION, DEPLOYMENT, AND IT OPERATIONS
Develop detailed plans for migration, including:
A comprehensive functional design
Systems and solutions for production, development, test, and staging environments Detailed migration timelines, schedules, and resources
Disaster recovery and failover solutions Proof-of-concept testing
Application migration, including tools and utilities Final deployment and go-live
Developing thorough and detailed cut-over plans
Sample Mainframe Migration Checklist
Selecting testing tools and procedures for:
Performance testing and infrastructure validation Stress and end-to-end system functionality testing Create plans for deployment and operations, including:
On-going support and maintenance
Integration with IT management, operational tools, and procedures On-going security and governance
ASSESSING AND MITIGATING RISK
Identify essential resources and potential risks to the success of the project, including:
Key stakeholders, collaborators, and resources
Internal and external team dependencies and requirements Developing contingency plans, including roll-back plan QUANTIFYING COSTS AND BENEFITS
Create detailed reports of benefits, costs, and risks, including:
Business and cost benefits, including license, systems, maintenance, and operational costs TCO, ROI, CAPEX, and OPEX analyses, including projected cost savings
Sample Mainframe Migration Checklist
For more details on the Intel Xeon processor family, visit www.intel.com/go/xeon
† Software and workloads used in performance tests may have been optimized for performance only on Intel microprocessors. Performance tests, such as SYSmark and MobileMark, are measured using specific computer systems, components, software, operations and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products. For more information go to http://www.intel.com/performance.
1 Source: HSBC
2 Server configurations for application server performance on the SPECjAppServer*2004 benchmark. SPECjAppServer is a trademark of the Standard Performance Evaluation Corp. (SPEC). Competitive numbers shown reflect results published on www.spec.org as of September 21, 2010. The comparison presented is based on the best single-node four-socket results. For the latest SPECjAppServer2004 results, visit www.spec.org/osg/jAppServer2004.
• Four-socket Intel® Xeon® processor X7350 (8 M cache, 2.93 GHz, 1066 MHz FSB): HP ProLiant* DL580 G5 platform with four Intel Xeon processors X7350 (8 M cache, 2.93 GHz,1066 MHz FSB), 65,536 MB memory, Red Hat Enterprise Linux* 5 Update 1 IA32 PAE, Oracle WebLogic* Server Standard Edition Release 10.3. Referenced as published at 3,339.94 SPECjAppServer2004 JOPS@Standard. Source:www.spec.org/osg/jAppServer2004/results/res2008q3/jAppServ- er2004-20080715-00105.html.
• Four-socket Intel® Xeon® processor X7460 (16 M cache, 2.66 GHz, 1066 MHz FSB): HP ProLiant DL580 G5 platform with four Intel Xeon processors X7460 (16 M cache, 2.66 GHz,1066 MHz FSB), 65,536 MB memory, Oracle Enterprise Linux 5 Update 2 x86_64, Oracle WebLogic Server Standard Edition Release 10.3. Referenced as published at 4,410.07 SPECjAppServer2004 JOPS@Standard. Source: www.spec.org/osg/jAppServer2004/results/res2008q3/jAppServ- er2004-20080826-00112.html.
• Four-socket Intel® Xeon® processor X7560 (24 M cache, 2.26 GHz, 6.40 GT/s Intel® QuickPath Interconnect (Intel® QPI)). Dell PowerEdge* R910 server platform with four Intel Xeon processors X7560 (24 M cache, 2.26 GHz, 6.4GT/s Intel QPI), Intel® Hyper-Threading Technology enabled, Intel® Turbo Boost Technology enabled, 131,072 MB memory, 2-disk SAS 72 GB 15K RAID-0 array, Oracle Enterprise Linux* 5 Update 4 x86_64. Oracle WebLogic Server Standard Edition Release 10.3.3. Referenced as published at 11,057 SPECjAppServer2004 JOPS@standard. Source: www.spec.org/osg/jAppServer2004/results/res2010q1/jAppServer2004-20100310-00140.html.
• Four-socket Intel Xeon processor X7560 (24 M cache, 2.26 GHz, 6.40 GT/s Intel QPI). Hewlett-Packard ProLiant DL980 G7 server platform with eight Intel Xeon processors X7560 (24 M cache, 2.26 GHz, 6.40 GT/s Intel QPI), Oracle Web- Logic Server Standard Edition Release 10.3.3. Referenced as published at 20,092 SPECjAppServer2004 JOPS@standard. Source:www.spec.org/osg/jAppServer2004/results/res2010q3/jAppServer2004-20100825-00145.html.
3 Server configurations for Java* performance on the SPECjbb*2005 benchmark. SPECjbb is a trademark of the Standard Performance Evaluation Corp. (SPEC). Comparison based on best published/submitted results on www.spec.org as of September 21, 2010.
• Four-socket Intel® Xeon® processor 7140M (16 M cache, 3.40 GHz, 800 MHz FSB): HP ProLiant ML570 G4 platform with four Intel Xeon processors 7140M (16 M cache, 3.40 GHz,1066 MHz FSB), 32 GB memory PC2-3200, Microsoft Windows Server* 2003 R2 Enterprise x64 Edition with SP1, Oracle JRockit* 5 P27.1.0. Referenced as published at 217,334 BOPS. Source:www.spec.org/osg/jbb2005/results/res2006q4/jbb2005-20061121-00222.html.
• Four-socket Intel Xeon processor X7350 (8 M cache, 2.93 GHz, 1066 MHz FSB): Dell PowerEdge* R900 server platform with four Intel Xeon processor X7350 (8 M cache, 2.93 GHz,1066 MHz FSB), 64 GB memory, Microsoft Windows Server 2003 Enterprise x64 Edition SP1, Oracle JRockit 6 P28.0.0. Referenced as published at 537,116 BOPS. Source:www.spec.org/osg/jbb2005/results/res2009q1/jbb2005-20090224-00592.html.
• Four-socket Intel Xeon processor X7460 (16 M cache, 2.66 GHz, 1066 MHz FSB): Fujitsu PRIMERGY* RX600 S4 server platform with four Intel Xeon processors X7460 (16 M cache, 2.66 GHz, 1066 MHz FSB), 64 GB memory, Microsoft Windows Server 2003 R2 Enterprise x64 Edition, Oracle JRockit 6 P28.0.0. Referenced as published at 633,897 BOPS. Source:www.spec.org/osg/jbb2005/results/res2009q1/jbb2005-20090305-00663.html.
• Four-socket Intel Xeon processor X7560 (24 M cache, 2.26 GHz, 6.40 GT/s Intel QPI): Cisco UCS* C460 M1 server platform with four Intel Xeon processors X7560 (24 M cache, 2.26 GHz, 6.40 GT/s Intel QPI), Intel Hyper-Threading Technology enabled, Intel Turbo Boost Technology enabled, 256 GB memory, 1x 73 GB disk drive, Microsoft Windows Server 2008 R2 Enterprise x64 Edition, IBM J9* JVM (build 2.4, JRE 1.6.0, SR5). Referenced as published at 2,021,535 SPECjbb2005 bops and 126,345 SPECjbb2005 bops/JVM. Source: www.spec.org/osg/jbb2005/results/res2010q2/jbb2005-20100331-00838.html.
• Eight-socket Intel Xeon processor X7560 (24 M cache, 2.26 GHz, 6.40 GT/s Intel QPI): Hewlett-Packard ProLiant DL980 G7 server platform with eight Intel Xeon processors X7560 (24 M cache, 2.26 GHz, 6.40GT/s Intel QPI), IBM J9 VM, Windows Server 2008 Enterprise SP1. Referenced as published score of 3,816,799 SPECjbb2005 bops and 119,275 bops/JVM. Source: www.spec.org/osg/jbb2005/results/res2010q3/jbb2005-20100816-00914.html.
• 32-socket Intel Xeon processor X7560 (24 M cache, 2.26 GHz, 6.40 GT/s Intel QPI): SGI Altix UV* 1000 server platform with 32x Intel Xeon processors X7560 (24 M cache, 2.26 GHz, 6.40 GT/s Intel QPI), 1,048,576MB memory, SUSE* Linux Enterprise Server 11 SP1, Oracle JRockit P28.0.0. Referenced as published score of 12,665,917 SPECjbb2005 bops and 98,952 bops/JVM. Source: www.spec.org/osg/jbb2005/results/res2010q3/jbb2005-20100616-00867.html.
INFORMATION IN THIS DOCUMENT IS PROVIDED IN CONNECTION WITH INTEL® PRODUCTS. NO LICENSE, EXPRESS OR IMPLIED, BY ESTOPPEL OR OTHERWISE, TO ANY INTELLECTUAL PROPERTY RIGHTS IS GRANTED BY THIS DOCUMENT. EXCEPT AS PROVIDED IN INTEL’S TERMS AND CONDITIONS OF SALE FOR SUCH PRODUCTS, INTEL ASSUMES NO LIABILITY WHATSOEVER, AND INTEL DISCLAIMS ANY EXPRESS OR IMPLIED WARRANTY, RELATING TO SALE AND/OR USE OF INTEL PRODUCTS INCLUDING LIABILITY OR WARRANTIES RELATING TO FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT. UNLESS OTHERWISE AGREED IN WRITING BY INTEL, THE INTEL PRODUCTS ARE NOT DESIGNED NOR INTENDED FOR ANY APPLICATION IN WHICH THE FAILURE OF THE INTEL PRODUCT COULD CREATE A SITUATION WHERE PERSONAL INJURY OR DEATH MAY OCCUR.
Intel may make changes to specifications and product descriptions at any time, without notice. Designers must not rely on the absence or characteristics of any features or instructions marked “reserved” or “undefined.”Intel reserves these for future definition and shall have no responsibility whatsoever for conflicts or incompatibilities arising from future changes to them. The information here is subject to change without notice. Do not finalize a design with this information. The products described in this document may contain design defects or errors known as errata which may cause the product to deviate from published specifications. Current characterized errata are available on request. Contact your local Intel sales office
