Platespin Orchestrate
The "brain" of the Data Centre automation system is Platespin Orchestrate (previously Novell ZENworks Orchestrator) This is the key component that interacts with configuration and storage resource management servers to manage physical compute and storage resources and the relationships between them. The Orchestrate also manages virtual resources, controlling the entire lifecycle of each virtual machine. The Orchestrate is a third-generation orchestration engine that is built from the ground up to meet the requirements of on-demand computing. It combines the following capabilities:
In general, competitive products feature a global job scheduler and workload manager to orchestrate "jobs" across a distributed environment. But to legitimately orchestrate the operations of today's complex, sophisticated data centres, Novell believes it is necessary to have competencies in:
- Resource management
- Job management
- Dynamic provisioning
- Policy management
- Accounting and auditing
- Real-time availability
Rather then working through in a detailed example of each of these areas, let us use an extremely simplified example of how only Orchestrator might manage our Data Centre.
Simple Example
Take a simple example - a Company that has 10 database servers that run an overnight application on 10 servers, and 10 servers that run the B2B web services application, primarily during the day. In a traditional environment this would require some 20 individual servers and the attendant storage, and infrastructure.
With ZENworks Orchestrator we can change this view...
By making use of the XEN hyporvisor (or ESX or Virtual Server and the network infrastructure we can create the same 20 servers, virtualised. During the daylight hours we run each of the virtualised Web Services on their own dedicated hardware with the "sleeping" database applications consolidated within a single piece of hardware
The requirement for the daylight application (Web Services) is between 08:00 and 18:00
The Night-time database process starts at 21:00 and lasts until 03:00 and we reverse the deployment with the Web Services consolidated (but available) on a single server.
And again we can see that each of the database servers has it's own physical hardware to run upon. In this simple example we have saved 45% of the hardware costs to provision this Data Centre with the attendant savings in the infrastructure, but there is one item that has been left out. The applications as specified run from 08:00 to 18:00 and from 21:00 to 04:00, what happens during the off hours when neither application is required? Well that is the key to Green Computing.
By controlling the servers centrally and getting them to power off when they are not required, instead of 20 servers sitting there idle - consuming power and creating heat, we now have just 2 - a 90% reduction - but without the services ever being unavailable.
In a full sized Data Centre, we are not talking in 10's of servers but 1,000s and with that, the attendant savings.
Cooperative Resource Pre-emption
A collection of jobs, all under the same hierarchy can cooperate with each other so that when one has offered to give up a resource it will be reallocated to another similar priority job.
Resource Stealing
When a job of higher priority becomes overloaded and is waiting on a resource, the system steals a resource, from a lower priority job, increasing load on the low priority job and allocates it to the higher priority job. Thereby satisfying the policy, which said a higher priority job must complete at the expense of a low priority job.