A global oil and gas company in the U.S. had three operational data centers - in Houston, Oklahoma City and Calgary. With these data centers being located in leased towers, they had an increased risk from facilities-related issues. Natural disasters in these areas also affected the operation of the data centers. The data centers themselves were running out of space and power as the leased spaces were not built to formal data center standards.

The company was also building a new corporate facility and needed to move out of the leased buildings without endangering the safety and sustainability certifications of the new facility. A high-level ROI study indicated savings of up to $40M over 20 years with the building of a new facility over upgrading existing facilities.

A new data center facility was built from the ground up as the Primary Data Center (PDC) for the company. This 20,000 square ft building was being built to a tier 3 data center standard and has the ability to withstand a F5 tornado or 315 mph winds. Highlights of this Leeds Certified data center include:

• 10,000 sq ft of initial data center space
• Up to five 9 availability
• 10,000 SF of future data center space
• Alternative Work Location (AWL) and office space for NOC
• Fully redundant mechanical, electrical and cooling systems( Evaporation Cooling/Air Cooling, Dual Electrical and Telecom feeds, Multi generator back ups )

IT Migration

The planning to move into this brand new data center was initiated in mid 2009 with December 2010 as the targeted date of completion. The standard process of asset mapping, application mapping, creation of affinity groups, migration planning and execution was followed throughout the life cycle of the project. This migration project has been immensely successful with the project fulfilling or exceeding all KPIs and Variance models. This article examines the keys to this success, opportunities that could have improved the delivery of this program as well as a look ahead on what remains to be strategically done.

The initial asset discovery data was a manual process executed through a web form and other spreadsheets from the support IT population. A base application list from the Enterprise taxonomy was used to populate and fine tune the asset data collected. Asset application mappings were created manually and affinity groups created from this data. The migration groups were designed along business lines to enable effective communication and lower impact to business during any migration event. About 2,000 assets and 1,600 applications were in the scope of the migration ,which was divided into 12 migration groups, six of which were nonproduction. The first of the migrations was done in March 2010 and the last in Oct 2010.

The success of this program can be attributed to many factors:

Enablers

1. Network: The new data center has a new core infrastructure and network build that is completely different than the existing network infrastructure with a new set of VLANs and IP Addresses. Early in the planning it was decided to extend the layer two network from the new Data Center to the old data center. Assets were migrated to this extended network in the old data center, stabilized if necessary and then moved to the new data center. This greatly reduced the risk of the actual migration.
2. Technology: Virtualization of the existing architecture prior to the migration was a great enabler. Pre-migration, the company's data center architecture was about 65% virtualized. The migration of virtualized servers through SRM for Intel and Zone transfers over the wire greatly reduced physical migrations and associated problems.
3. Methodology: The actual migration planning for each move was a repetitive process of moving each group of assets through the same process over a period of eight weeks. After the initial moves, this repetition removed any associated surprises and the actual moves went without issues.
4. Communication: The greatest enabler was the establishment of a repetitive communication method to the organization about upcoming moves and events. A weekly workshop with the BAs and SAs involved in the move from eight weeks prior; a repetitive communication to the Business stakeholders before every move; and the company wide communication about the migrations were instrumental in controlling anxiety in the business as well as giving the business time to prepare for the moves.
5. Resources: Ample seed hardware was procured and set up in the new data center in terms of Wintel and Solaris servers. These servers acted as migration hosts for virtual servers as well as were used to remove old out-of-maintenance hardware from the DC. The human resources used were from within the company which aided communication and reduced ramp up times and technology assumptions from contracted resources.

Constraints

Analyzing the constraints which affected the program delivery:

1. Asset and Application Mapping: The manual process of asset and application collection leaves gaps in the data collected, The use of an automated tool is highly recommended for the collection of a comprehensive asset and application data set. This enables clear decision making on affinity groups and assets and applications do not fall through the cracks.
2. BCP: The availability of a mature business continuity plan aids migration decisions on the identified business critical applications. Migration of business critical applications, the involved downtimes etc can be better planned than for non critical applications and devices.
3. Testing Tools: Making a testing sandbox available for testing of applications in the development environment goes a long way in alleviating anxiety in the BA/SA community, It also gives the migration team the ability to identify applications that can have migration related problems and mitigate per business need.
4. Performance Testing and Benchmarking: The lack of performance testing and benchmarking information for applications leaves the interpretation of performance post migration open to human interpretation and is not data driven. Performance testing as well as benchmarking data is highly recommended as migrations of applications in affinity groups are being planned.
5. Executive Communication and Coverage: Constant communication from the executive team enables the communication to be more effective as departmental leaders become engaged much sooner and the communication flows both ways. Without this coverage, the communication is only one way, outwards from the migration team which may not be the most efficient.

What's Next?

The the assets of the company have now been migrated to the new data center. The new data center is one of the best-in-breed and has been Leed-certified in terms of sustainability and redundant systems. The following post-migration activities are recommended:

1. Data Center Optimization: The Data Center row and rack load need to be optimized in terms of power, cooling and business requirements(Hot/Cold Rows, HPC/HA rows, Dev,Stg/Prod rows ). This has the potential of reducing in row cooling and power loads and lower consumption in the Data center.
2. DR and BCP Review: Post the migration the Disaster Recovery Plan and the Business Continuity plan need to be reviewed and updated as the enterprise architecture has changed from the pre-migration days. This will involve reviewing and updating the HA strategy as well as the disaster recovery strategies in place.
3. Audits: With the new data center comes a new way of servicing applications. The new network and application infrastructure need to be re-certified as being Cobit, PCI-SOX and ISO 27002 compliant. These audits will enable the identification of opportunities to solidify the network and the infrastructure to be in line with industry security standards.
4. TCO and ROI: Now that the data center has been built as well as the migration into it completed, a validation of the ROI initially computed and the Total Cost of Ownership in owning and maintaining this data center is recommended. This will enable strategy decisions on the future of this state of the art data center and its use by the company
5. Technology Optimization: This has two parts to it. Firstly in the course of migration certain legacy applications were identified which have performance related issues when serviced over a LAN/WAN. Alternative applications/ or applications delivery methods need to be examined. Secondly with the advent of cloud computing, the company needs to leverage this maturing technology option to reduce costs of using/maintaining/owning a large data center of this sort.

With the advent of cloud computing strategies and the movement of many smaller enterprises to the cloud, this large data center can be leveraged to improve ROI or lower the cost of ownership by the leadership in the company. This is a situation many CXOs will be facing in the next decade as cloud computing matures and becomes the norm rather than the exception.