Data center technology

The advanced technologies in our Data Center provide many features our customers can benefit from to have a high level of scalability and resilience.

Dual Homing
Dual Homing allows the server to be connected to two different switches allowing resilience for the NIC. Dual homing is optional for both front-end and back-end servers

Clustering
Dual Homing allows the server to be connected to two different switches allowing resilience for the NIC. Dual homing is optional for both front-end and back-end servers

Clustering

Clustering provides a fault tolerant, highly available bank of machines. All machines contained within the cluster virtually appear as one machine to the end-user. Clustering ensures that the critical applications are online when needed by removing the physical server as a single point of failure. In the event of a hardware or software failure in either node, the applications currently running on that node are then migrated to the surviving node and restarted.

At Raya Telecom, the servers are configured as Active-Active. Meaning that the two servers are always working together at any given time to run the clustered application. In case one of the servers fail, the other one takes over. The active-active configuration increases availability as well as performance.

Cluster service is ideal for ensuring the availability of back-end systems, such as Microsoft Exchange Server or a database server acting as a data store for an e-commerce Web site.

Load Balancing

Network Load Balancing (NLB), included with Windows 2000 Advanced Server, can be used to balance incoming Internet traffic across a farm of servers. When Network load balancing is installed as a network driver on each of the cluster hosts, the cluster presents a virtual IP address to client requests. The client requests go to all hosts in the cluster, but only the host to which a given client request is mapped accepts and handles the request. All the other hosts drop the request.
 
A statistical mapping algorithm is present on all hosts in the cluster. The algorithm takes into account the port rules and affinity settings, and maps the client request to a single host for processing. The hosts exchange heartbeat messages to maintain consistent data about the cluster’s membership. If a host fails to send or does not respond to heartbeat messages, the remaining hosts perform convergence. Through convergence, the hosts determine which hosts are still active members of the cluster.
 
 When a new host joins the cluster, it sends heartbeat messages that trigger convergence. After all cluster hosts agree on the current cluster membership, the client load is repartitioned and convergence completes.

The scalability and load balancing occurs by virtue of the load balancer directing resource requests across the front-end servers to best balance the load for the site. The load-balancing algorithm determines which server actually responds to a user request.

When the traffic on the site increases beyond the capacity of the site, new Web and Application servers can be configured with the Web content and components. When the new Web and Application server is booted up on the network, it will dynamically join the existing group of Web and Application servers and immediately begin sharing the load.

SAN Storage

SAN devices are essentially large storage buckets. Storage Area Networks (SANs) allow systems administrators to collect large amounts of disk storage in highly localized and manageable units, such as large servers full of RAID arrays. Application servers and users can then store their data on these large bit buckets. The file system runs on the application server and uses block level I/O to access SAN devices. SANs typically use fiber channel and fiber switches to connect to application servers.