Reference architecture -- starting point to build and optimize cloud infrastructure

Rekha Raghu, Strategic Program Manager, Intel, Software and Services Group, discussed some reference architecture (RA) case studies.

Intel Cloud Builders program is a reference architecture (RA) -- a starting point from where to build and optimize cloud infrastructure. The RA Development Process takes anywhere from two to three weeks. It involves exploration, planning, integration, testing and development. The RA is said to be:
* Detailed know-how guides.
* Practical guidance for building and enhancing cloud infrastructure.
* Best-known methods learned through hands-on lab work.

RA case study # 1 – efficient power management
Data center power management involves monitor and control server power, and later, manage and co-ordinate at data center level. Dynamic power management is on the server, rack, and data center levels.

Power management use cases help to save money via Real time monitoring, optimized workloads and energy reduction. They allow scaling farther via power guard rail and optimization of rack density. They also help prepare for the worst in terms of disaster recovery/business continuity.

Intel also presented a power management RA overview as well as an implementation view. The monitoring, reporting and analysis provides insight into energy use and efficiency, as well as CO2 emissions.

Rack density optimization and power guard rail enables more servers deployed per rack. It improves the opex cost of power delivery per system. It also extends the capex data center investment with increased node deployments.

As for disaster recovery/business continuity, there is policy based power throttling per node to bring the data center back to life more quickly and safely. The next step involves inlet temperature monitoring and response based on thermal events (already available in Intel Intelligent Power Node Manager).

Workload-power optimization identifies optimal power reduction without performance impact. Customized analysis is required as each workload draws power differently.

RA case study # 2 – enhanced cloud security
If one looks at the trends in security in the enterprise, there are shifts in types of attacks. The platform is now as a target, not just software. Stealth and control are taken as objectives.

There are increased compliance concerns. HIPPA, Payment Card Industry (PCI), etc., require security enforcement and auditing. Changes in architectures require new protections as well. These include Virtualization and multi-tenancy, third party dependencies, and location identification.

Trustable compute pools usage models lead to compliance and trust in the cloud. Compliance in the cloud involves multi-tenancy that could complicate compliance. There is need for software trust despite physical abstraction. Also, compliance requires effective reporting. There is a need to enforce VM migration based on security policy.

Intel-VMware-HyTrust enables trusted compute pools. The outcome is that data integrity is secure and there is no compliance violation.

Intel Trusted Execution Technology (TXT) enforces platform control. It allows greater control of launch stack and enables isolation in boot process. It also complements runtime protections, and reduces support and remediation costs. Hardware based trust provides verification useful in compliance.

HyTrust appliance enforces policy. It is a virtual appliance that provides unified access control, policy enforcement, and audit-quality logging for the administration of virtual infrastructure.

Intel provides solutions to pro-actively control and audit virtualized data centers.