This is a complex task with challenging customer requirements, but the idea is to collapse the existing equipment at 150 locations to a small number of data centres with full redundancy. The new architecture centralises all data for analytics and reporting purposes, while continuing to support the varying requirements of around 300 applications.
Part of the project has been to determine exactly what is in the Defence IT environment and identify all the dependencies (for example, which applications are affected if a particular network switch fails).
It has provided an opportunity to simplify and standardise the environment, so a much smaller range of operating systems, databases and so on need to be maintained.
The new infrastructure allows Lockheed Martin to provide appropriate service classes to suit various applications, and to see immediately when anything goes wrong, allowing prompt remedial action. Trouble tickets are raised automatically when a component fails.
"I love automation," Caskie told iTWire. Other examples of automation in the new private cloud include virtual machine setup, load distribution across VMware clusters, and disaster recovery - whether that means failing over to another data centre or just to another component at the same location.
Caskie's goal is to automate everything from build to operations.
The company is acting as a managed service provider to the Defence Department, and as such provides its customer with a service level dashboard that sits on top of all the other components. The idea is that the dashboard should save Defence staff from having to burrow into check that performance standards are being met (eg, that trouble tickets are actually being fixed) or to check that recommendations for the supply of additional resources such as extra storage capacity are justified.
The project has been underway for about a year and is still in progress. Getting the various application owners onboard with the move to private cloud has been a challenge because some of the applications - notably those that directly support military operations - are extremely critical.
There were three main stages: remediation (typically work needed to ensure that the applications were compatible with the latest versions of system software such as operating systems and databases), rehosting (moving systems onto cloud-based virtual machines) and retirement (the project revealed that around 12% of the department's applications were no longer required, which provided a significant saving by reducing the hardware needed).
While there have been some software issues, they are "now well and truly licked," Caskie said, and the systems should go online at the end of this month (October 2015).
A second release is planned for November, and then a third will complete the bulk of the required functionality. This phased approach "turns a massive milestone into achievable chunks," he told iTWire.
The development work - from application remediation to handle the cloud environment as well as the integration effort - has been performed by Lockheed Martin's own staff in Melbourne.
The Defence Department was already a NetApp customer, and using NetApp-accredited designs saved time during the project, said Caskie. Furthermore, the relationship with NetApp is "sensational," he said, and access to the company's experts does not always cost extra.
Some EMC equipment is used for archiving, but using 100% NetApp products for primary storage simplifies the architecture. "I'm looking to get as much computing out of the environment as possible," he said, noting that minor hardware upgrades such as the installation of more network ports meant existing arrays could be kept in service in the private cloud instead of spending a lot more on replacement units.
Caskie is looking forward to reducing his hands-on involvement once the private cloud rollout is largely complete, as that will allow him to spend more time developing the department's IT strategy.
Disclosure: The writer attended NetApp Insight as a guest of the company.