Remsys logo

Cloud computing and clusters deployment

 
Customers expect their servers to be fast, up and running 24x7.
They also expect scalable servers that transparently meet growing processing power demand.
Deploying larger servers to meet these customers expectations is an expensive, short-term approach, passible to eventual failure.
Actually providers  are finding that combining load balancing with low-cost clusters, high-density applications servers is a far more cost-effective, flexible, robust and reliable strategy.

What are server clusters?
A cluster is a group of physical servers with identical content, networked together to form a single virtual server (VIP).
Requests directed to the VIP can be handled by any physical server in the cluster.

  • clustering enables transparent growth - physical servers can be added without externally-visible end-user changes.
  • clustering improves fault-tolerance - a physical server can be turned off for maintenance or repair without resulting in a service outage.
  • a server cluster could provide high-availability and throughput characteristics far beyond those achievable by even the largest single server.

What is Load Balancing?
DNS balancing could be used to redirect requests to different servers in a round-robin way.
Hovewer simple round-robin DNS cannot find more powerfull servers, react to real-time changing conditions, or avoid unavailable servers.
For a cluster of servers to achieve its high-availability and high-performance potential, load balancing is required.
Load balancing optimizes request distribution based on multiple factors like processing power, actual load, availability, response time,  and administrative weights.
A load balancer sits between the external network and a physical server cluster, acting as a VIP.
As each request arrives, the load balancer makes instantaneous intelligent decisions about the physical server best able to satisfy the incoming request.
A well-tuned load balancer ensures that customer applications are available 24x7 with the best possible response time and resource utilization.

REMSYS has experience implementing following types of clusters :

  • Web clusters
  • Mail clusters
  • DNS clusters
  • Database clusters