February 6, 2013 | Written by: Kohhei Nomura
Share this post:
In my previous post, I wrote an overview of Flexible Private Cloud Architecture (FPCA). In this blog, I would like to show you details of FPCA based on the figure below.
Service portal layer
Service portal acts as a one-stop portal in a private cloud environment. It provides a user-friendly interface for resource reservation and service desk related inquiries. Also it provides several workflows in order to handle various application processes without fail. Users, approvers and operators access the service portal and carry out entire utilization processes.
Automation tool layer
The automation tool integrates multiple resource pools and it acts as a single console for operators, thus the operator or service portal can manage different types of resource pools transparently. It provides automation functions such as provisioning and de-provisioning of virtual machines in resource pools. Also most of automation tools expose API which enables the service portal or any user programs to utilize automation function directly.
Nowadays most automation tools provide API with REST or web service. With FPCA, all user requests are transmitted through the automation API by controller layer, thus we can handle all user requests without having any human interactions. In a general private cloud environment, reducing human operation task results in reducing operational cost broadly.
Resource pool layer
Resource pool provides computing resources such as CPU, memory and disk for virtual machines. It is composed from physical servers, storage systems, networks and hypervisors. If necessary, we can divide the resource pool based on hypervisor type (PowerVM, VMware, KVM, Hyper-V) as written on the figure. In addition, we can also divide the resource pool based on system purpose (Enterprise System, Informational System, Office Automation) or service level (Gold, Silver, Bronze), for instance.
Management tool layer
From the perspective of cloud infrastructure management, it is important not only to monitor the near real-time system date, but also to store those real-time system data and to analyze resource utilization trend for future capacity planning.
Other important functions which the management tool layer needs to provide are as follows:
- Integrated console for monitoring multi-platform and multi-hypervisor environment
- Mapping of virtual servers and host
- Automated action based on threshold
A controller layer is a characteristic component in FPCA. It allows loose coupling of the service portal layer and automation tool layer. Consequently, if we want to change one automation tool to the other, we can change it without having any affect to the service portal.
With FPCA, service portal delegates processing to the controller at the timing of workflow progress by calling REST API, which the controller exposes. Subsequently, the controller calls automation API based on the request which the service portal transmitted.
A merit of using the controller is not only in reducing operational costs but also in providing simplicity to the service portal layer. Think about deploying a new virtual environment; it is necessary to call multiple APIs such as checking resource availability, creating project, creating virtual machines and more. With a controller, service portals do not have to call each of those APIs. Instead, the service portal only requests the deployment of new virtual environments to the controller and the rest of the tasks are handled by the controller layer.
For my next blog post, I’m going to show you how FPCA provides flexibility to a private cloud environment and refer to the “controller” which is a keystone of FPCA.