Migrate to the Future: Mitigate risks and reduce ... - IBM to the Future: Mitigate risks and reduce costs with ... –C/C++ application analysis ... IBM and Red Hat The Strategic Migration
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Are the required resources available?How will the existing systems administration and application development team skills be transitioned?How will retraining be performed?
Skills and Culture
How well will it work?How will the migrated workload be tested?Will the performance and reliability meet business requirements?Will it work the same way on the target platform or will changes in customer, supplier or user interfaces be required?
Operational
Can it be done on time?How much downtime will be required for transition?When can the business support this change?
Schedule
Can it be done within the budget?How will the migration cost be funded?Does the business case have a positive ROI?
Cost
Can it be done?Are required ISV products available on the target platform?What differences need to be addressed such as application APIs, threading and data formats?Are there tools available to help minimize the complexity and risk?
To help minimize the cost of transition/ migration services so they do not become the main objection to moving to IBM–The Migration Factory’s core competencies–Provide and leverage many person-years of
application migration experience–Mitigate and reduce the risk in moving
applications from one platform to another–Reduce the cost of moving applications from one
platform to another–Support success through process, expertise and
• Print and file management• Web• Security/firewall• Technical• Application development• Other
Complexity and time drive migration costs
Moving HP and Sun UNIX workloads to IBM involves four basic types of migration, with varying levels of difficulty.Oracle database migrations are frequently a starting point to introduce IBM hardware.
Utilize tools wherever possible — standard UNIX tools, commercial tools and our own custom toolsReuse code where ever possible, recode only when necessaryFor custom applications migrate like for like, bugs as well!Code synchronizations available for new development and defect fixesMigrating new applications in development involves too much riskISV applications and databases can be upgraded to new versions as part of the migration process
1. Solaris vs. RHEL Ecosystem Analysis – Mapping the Solaris ecosystem into the RHEL ecosystem and creating a RHEL Standard Operating Environment (SOE) .
2. Functional Applications Analysis – High level analysis of business applications to be migrated.
3. Organizational Readiness and Risk Analysis – Analysis of organizational readiness factors, project risks, and risk mitigation strategies.
4. Strategic Migration Roadmap Creation – Combining everything into a single, holistic roadmap for migration.
5. Migration Implementation – Execution of the Strategic Migration Roadmap.
Examine existing Solaris ecosystem and determine the equivalent capabilities in the RHEL ecosystem.Create a gap analysis and plan to address all gaps (if needed).Create a Standard Operating Environment (SOE), an organization's standard implementation of RHEL, including base operating system, a custom configuration, standard applications, software updates and service packs.
Analyze complexity and size of existing functional applications to determine macro-level migration difficulty.Analyze application migration dependencies, including tightly coupled interfaces and co-resident applications. Examine possible deployment scenarios for each application and its associated testing and staging environments based on the four generic deployment patterns.Create high-level functional migration application cost analysis.
Phase 3: Organizational Readiness and Risk Analysis
Examine organizational readiness factors including skill gaps, IT governance processes, and acceptance factors.Perform situational SWOT analysis to determine current-state migration strengths and weaknesses as well as future opportunities and threats.Analyze technical risks inherent in many migrations.Create Risk Mitigation Strategy to address and limit the impact of identified risks.
Create final list of application workloads to be migrated.Perform consolidated server, deployment, and virtualization analysis.Examine hardware redeployment scenarios and opportunities.Create detailed training plan to address all identified skill gaps.Create detailed direct cost estimate for the entire migration.Create master Migration Roadmap.
Development environmentKernel tuningSecurityFilesystemsDebugging, tracing, profilingSoftware managementVirtualization3rd party application considerations
Development environmentCode Compilation– Differences in compilers & flags (gcc vs. cc) – Versions of make (make, nmake, gmake) – Linker option differences– System-specific APIs such as Solaris “doors” can
cause issuesEndianness – the “NUXI” problem– Systems can be “big-endian”, “middle-endian”, or
“little-endian”– A date analogy:
• US date format is middle-endian: 10/06/2009• EU date format is little-endian: 06/10/2009• Big-endian date format would be: 2009/10/06
– For the most part, SPARC, PA-RISC, and IBM Power are big-endian
– Linux / x86 is little-endian– Can cause lots of problems, particularly in programs
In many cases, RHEL’s kernel can be tuned dynamically without rebootingIn Solaris, kernel changes are done in /etc/system and require a reboot to take effectSome instant changes are possible in Solaris through tools such as ndd, adb, DtraceIn Linux, pseudo-filesystem /proc is the kernel's memory, available for instant modificationChanges can persist reboots in /etc/sysctl.conf
BART-like functionality available through AIDE (Advanced Intrusion Detection Environment) Tripwire also availableSecurity Enhanced Linux (SELinux) co-developed by the NSA is included with all versions of RHELNo separate “insecure” version of RHEL (Solaris vs. “Secure Solaris”) Common Criteria/EAL4+ certificationServices are disabled by default even after installed or configured to listen to localhost
Debugging, tracing, profilingSolaris – Dtrace is very popular and powerful
RHEL and Linux community use SystemTap and Oprofile – RH has core developers on both teams
SystemTap:– Scripting language with full control structures– Millions of probe points in kernel & user spaces– Probe arbitrary statements in code– Symbolically extract arbitrary data at probe point– End-user extendable probe library (tapsets) – Protected probe execution environment– Division by zero, null pointer, infinite loop protection
Oprofile:– System-wide profiling of all running code– Unobtrusive, low overhead, no recompilation– Post-profile analysis
Kernel Virtual Machine uses Intel's VT-x CPU extensions to offload virtualization tasks to the CPUEffectively provides new privilege levelRemoves need for hypervisor to scan and rewrite kernel codeExtended Page Tables in NehalemI/O Offloading (VT-D) – PCI Pass-throughSingle Root I/O Virtualization (SR/IOV) –Split PCI devices into virtual onesUses mature, stable, and proven kernelSame security and performance as RHEL
Barclays Bank New York Life BellSouth AT&T Mobility
La Quinta (Informix 4GL)Bloomberg (C/C++)The Home Depot (Informix 4GL/Java/C++)Dow Jones (C/C++/Sybase)Reuters (C/C++/Java)
VerispanDollar Thrifty (Oracle)IVAX (Oracle)Portland General Electric (Oracle)Wells Dairy (Oracle)Pitney Bowes (Oracle)Polo Ralph Lauren (Oracle)Wellington Management (Oracle)
Baylor College of Medicine (SAP)Invensys (SAP)Key Bank (PeopleSoft)Randstad(PeopleSoft)PetSmart (SAP)Pratt & Whitney (SAP) Philips Medical (SAP)Osram Sylvania (SAP)Adecco (PeopleSoft/OEBS)