Browse Prior Art Database

Method to automatically detect machine type and provision the OS with a machine-specific profile Disclosure Number: IPCOM000198935D
Publication Date: 2010-Aug-17
Document File: 2 page(s) / 20K

Publishing Venue

The Prior Art Database


Disclosed is a system and method to enable automatic detection of the machine type and various other hardware characteristics of the machine in order to make the decisions for servers about which OS or VM image, as well as other boot parameters, to boot.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 1 of 2

Method to automatically detect machine type and provision the OS with a machine-specific profile

In a large-scale cluster/cloud environment, where the server hardware is managed by a central management infrastructure, a considerable amount of time is spent in the initial "bring-up" of the servers. Assuming the new servers are attached to the management network, they need to be first discovered by the management software, and then the operating system or virtual machine (VM) image needs to be provisioned on the servers over the network using a network install or boot technique.

Typically, in a cluster/cloud environment today, the decision to deploy a particular operating system (OS) image or VM image on a particular server is made more or less statically: for a particular new machine type that is anticipated to be added to the cluster, its characteristics such as the machine architecture, number and type of processors, amount of physical memory, number of disk drives, type of network interfaces, etc., are either known beforehand or certain generic assumptions of these characteristics are made a-priori and the OS or VM image is predefined for that particular server based on this prior knowledge or the generic assumptions.

For example, assume that if a machine is known to have 48GB of physical memory, then based on the benchmarking performed, up to 12 Virtual machines can be launched on this physical server. Hence, any time a new server with 48GB memory is added to the cluster, the management software can deploy up to 12 virtual machines on this new server, provided that the knowledge about the presence of the 48GB memory in the server is known beforehand and the administrator of the management software can categorize this machine to have 48GB memory in the VM deployment software manually. Hence, the management software will be configured to boot up to 12 VMs on the server when it is discovered on the network.

The problem being addressed by this disclosure is how to dynamically make the decisions about which OS or VM image to boot as well as other boot parameters for servers. The goal is to detect the physical characteristics and other attributes of the server when the server is initially discovered and then make a decision as to which OS image or VM image to deploy on the server. Doing this determination dynamically versus statically enables a great deal of automation when managing a heterogeneous environment of servers and application workloads that require systems to dynamically deploy a variety of OS or VM images on the servers based upon specific characteristics of servers.

The main idea of this invention is to enable automatic detection of the machine type a...