Edge computing is evolving into the most preferred architecture for running data-driven, intelligent applications. As the push towards centralized computing continues, it is becoming apparent that all data cannot and will not be moved to the central cloud but will rather live at the edge. The new edge layer is the digital workspace. The amount of data generated at the edge is expected to be almost 2 orders of magnitude greater than what can be moved to the cloud. A number of architectural efforts are underway to address the needs of the distributed data processing paradigm with the growth of IOT, VR, AI and real-time decision making systems. This is expected to be long term effort with a pragmatic near-term approach that the 8kpc architecture is designed to support. The key requirements for this approach are two fold:
- Centrally managed resources: Even though the actual physically infrastructure (and data) is located in multiple places, the multiple resource pools (clouds) are managed centrally. The 8kpc platform with central orchestration and instrumentation enables the entire application deployment to be managed as a single virtual deployment.
- Central application orchestration: From a application perspective, the ability to manage all deployments from one place, managing the application life-cycle with a single view significantly reduces the burden of the IT resources while providing a uniform experience across the distributed enterprise.
Centrally managed resources
Whether a large enterprise with a number of branches or a service provider with multiple customers (tenants), the ability to manage distributed deployment from a single place significantly reduces the operating costs. The 8kpc platform supports the central management paradigm with the following key features:
- Single dashboard to manage multiple clusters that can run on any public or private cloud.
- Message bus based data collection of remote hardware clusters for performance and anomaly detection and monitoring.
- New resource deployment (new appliances) and scaling with central orchestration.
Central application orchestration
The key to reducing the time required to bring up new applications in a complex environment is eliminating the process of configuring and tuning the hardware infrastructure and the network. 8kpc’s approach is to understand applications at a deep level in the labs and create templates (covering multiple layers) that greatly simplifies configuring the components at each deployment location. This application specific intelligence is embedded in the 8kpc platform enabling simple remote orchestration of the applications. The key attributes related to central application orchestration are:
- Ability to analyze and understand the signatures of complex application deployments from data movement, data storage and resource consumption perspective (CPU, memory etc.).
- Optimize components tuned to application needs
- Centralized “app-store” construct to simplify deployment and life cycle management of applications
- Capture complex application deployments including network configurations
End result is that with 8kpc, edge solutions can be deployed optimally with resilience and simplicity (even at large scale) at a low cost.