AMRAPALI INSTITUTE
Vinay Kumar
5/6/2013
CSE 4th Year
1
Introduction.
Cluster and Supercomputers. Cluster Types and Advantages.
Cluster Performance.
5/6/2013
Group of computers and servers (connected together)
that act like a single system.
Each system is called a Node.
Node contain one or more Processor , Ram ,Hard disk and
LAN card.
Nodes work in Parallel. We can increase performance by adding more Nodes.
5/6/2013 3
5/6/2013
A supercomputer is a computer that performs at or
near the currently highest operational rate for computers.
A supercomputer is typically used for scientific and
engineering applications that must handle very large databases or do a great amount of computation.
IBM's Roadrunner is the fastest supercomputer in the
world.
5/6/2013 5
5/6/2013
5/6/2013
Load Balancing Cluster ( High Performance Cluster).
Visualization Clusters.
Grid Computing.
5/6/2013
It increases computer performance exponentially by
sharing the workload, spreading any given task among different cluster nodes.
serial and parallel.
The two main types of jobs that are run on HPC clusters are
Parallel jobs break the problem up into parts and use a
communication layer to spread the workload among many machines and processors. to other nodes.
A serial job runs on one machine with no communication
5/6/2013
Cluster types: Load Balancing Cluster
Task
5/6/2013
10
Running Program(sequential)
Working
5/6/2013
11
Running Program(sequential)
Working
5/6/2013
12
Running Program(sequential)
Working
5/6/2013
13
Running Program(sequential)
5/6/2013
14
Running Program(Parallel)
Data sent
Data sent
Data sent
5/6/2013
15
Running Program(Parallel)
Working Working
Working
Working
5/6/2013
16
Running Program(Parallel)
Finished Results Get results Results
Finished
Finished
Results
5/6/2013
17
A visualization cluster is an HPC cluster with addition of
powerful graphics cards.
Used to tackle high-resolution and real-time simulations. Weather forecasting DNA sequence analysis Climate and environmental modeling
5/6/2013
18
Geographical overview using Dells LongHorn Visualization Cluster.
5/6/2013
19
A grid ties many types of resources together in
differing locations.
These resources can be entire clusters. A grid allows an organization to pool multiple
computing resources: HPC clusters, visualization clusters, workstation and instruments.
5/6/2013
20
The Control Node sends the tasks to other computers that are connected to each other through an internetwork.
5/6/2013
21
Performance.
Scalability. Maintenance.
Cost.
5/6/2013
22
The performance of a cluster depends on the
performance of the systems in the cluster and the network that connects them.
Monitoring cluster performance requires a scalable
monitoring solution that integrates network and system monitoring.
5/6/2013
23
5/6/2013
24