Traditional datacenters are famously outdated and ill-equipped to handle the future. Sluggish and unpredictable, they are perceived to be unresponsive to the needs of the businesses they serve. They were designed as silos of excellence. Compute, storage, and networking infrastructure was segregated into separate domains, each separately procured, scaled, and updated, with no cross-collaboration. Lack of transparency meant stranded resources and compliance challenges.
The result was an unwieldy juggernaut that seemed to have lost sight of why datacenters exist in the first place: to provide the best experience for applications, the lifeblood of businesses.
Hyperscale infrastructure changes the game
The digital giants have changed it all with hyperscaled infrastructure, an architectural approach that combines cloud economics with software-defined infrastructure (SDI) principles. The promise is a fully-automated, tunable data center. Infrastructure is abstracted, pooled, and provisioned on demand.
Assembled using infrastructure building blocks, the hyperscale datacenter has programmatic control and line-of-sight across all datacenter assets – physical and virtual, corporate, and remote. In other words, there is full transparency of all assets – their capacity, performance level, utilization, and costs. And there is resource fluidity. Resources can be allocated and decommissioned as required in response to business priorities and service level objectives. With hyperscale everyone plays the role that makes most sense. Applications take center stage, and the infrastructure adapts to their needs.
Re-cast RFIs to realize the benefits of hyperscale
Proven in some of the industry’s most demanding and dynamic environments, hyperscale has evolved into a global movement. Now synonymous with datacenter modernization, companies everywhere are deploying hyperscaled data centers as the foundation for digital transformation.
To realize the full benefits of hyperscale, you may want to re-cast RFIs and map them to CxO KPIs and business outcomes. The following are some questions that you may want to ask your hyperscale vendors:
Real-time visibility across the datacenter:
- Do you provide a real-time view of the application infrastructure?
- Does your solution provide both visibility at the application portfolio-wide level as well as a granular workload-specific view of recourse utilization across the datacenter?
Dynamic resource allocation:
- Can compute, network, and storage resources be dynamically allocated without having to physically rewire servers and racks or change IP addresses to enable network services?
- Can infrastructure resources be specified, allocated, and measured at an individual application level?
- How rapidly can operations and infrastructure teams identify, prevent, and predict service-impacting bottlenecks and system failures?
- How quickly can your solution adapt to changing traffic patterns, including upstream data flows?
- What instrumentation and analytics capability do you provide? Do you support closed-feedback-loop to allow continuous tuning?
- How long does it take to scale your application infrastructure to adjust to unpredictable demand?
- Can storage resources be provisioned at the same time as compute and network resources?
Next generation applications and new application architectures:
- Do you support Tier 1 applications and workloads with each layer being high performance and low latency?
- Do you enable agile infrastructure and DevOps?
- Do you provide tools to alert operators to infrastructure bottlenecks, and to forecast and predict potential hotspots and recommend changes to the allocation of specific resources?
- Does your solution accelerate the workload lifecycle from prototyping and testing to deployment in production environments?
- Can your solution enable applications to be tested against the overall system architecture without disrupting day-to-day IT operations?
- How long does it take to provision an application with critical services, deploy it in production environments, and scale it on demand?
- Do you support bare metal and containerized workloads, in addition to virtual machines?
Infrastructure and IT operations automation:
- Do you provide a single, unified portal and an API-driven interface to operators for managing the infrastructure without the need to depend on CLIs, terminals, scripting?
- How open is your platform to enable integration of IT tools from ecosystem partners?
- Do you have application-specific blueprints to manage Day 2 Operations including reducing unplanned downtime and maintenance window, business continuity, and disaster recovery?
- Do you have a policy mechanism to express performance objectives and measure service availability and response levels?
Security and compliance
- Do you have a policy framework and profiles to dynamically reconfigure and reallocate application infrastructure to secure IT assets from being compromised?
- Do you have a policy framework to ensure business and regulatory compliance?
You can also speak to your local Intel cloud representative for more advice. Ericsson can help direct you as well.
To take a deeper dive, you can read our solution brief or explore our vision of Future Digital Infrastructure in a new paper based on a futuristic robotic datacenter we demonstrated at Mobile World Congress:
Ericsson believes in creating a new open way of doing digital infrastructure that will meet tomorrow's needs rather than accepting yesterday's compromises. 13 major IT infrastructure providers have already announced adoption of systems based on Intel® Rack Scale Design. Since our launch, Ericsson has 26 Hyperscale Datacenter System 8000 customers in all geographic areas.