Current Filter: >>>>>>

   

Many organisations run multiple instances of applications across globally distributed data centres to ensure high availability, optimise performance and reduce delivery costs. To maximise the benefits of this distributed infrastructure organisations need a mechanism to intelligently direct user requests to distributed resources and applications. The solution to this is called Global Server Load Balancing (GSLB).

GSLB is appropriate for a wide variety of use cases. Many organisations use GSLB for Disaster Recovery (DR) across applications in two or more Data Centres (DCs). GSLB solutions can perform health checks against application instances running in primary or backup DCs. When an application in a primary DC is running, all requests get directed to this primary DC. In case of application failure in the primary DC, all requests for this application will be redirected to the backup DC.

GSLB is also used by organisations to load balance user application requests across two or more active DCs. Doing so helps organisations to optimise performance and reduce application delivery costs.

GSLB offers a wide range of options to spread user demands across multiple servers and, or, data centres. Here are some of the most common:

• Global availability: Clients are directed to the first resource in a list of available resources. Only if the first resource becomes unavailable will GSLB direct clients to the next resource in the list
• Proximity: GSLB may use predefined geographic mapping and other user-defined mapping to direct the client to the closest available resource
• Ratio: GSLB will distribute requests among all or selected application instances based on a predefined ratio setting. For instance, 30 per cent of all requests may get directed to application instances located in the London data centre and 70 per cent of the requests will get directed to application instances located in the New York data centre
• Round Robin: A simple round robin algorithm may be used to distribute requests among available resources. For example, GSLB may be configured to direct the first requests to the London data centre, the second request to the New York data centre, the third request to the London data centre, and so on
• Best Performing: Some GSLB solutions have visibility into the performance status of an individual or a pool of applications. For example, a GSLB solution may be aware that applications hosted in the London data centre are running at a 70 per cent utilisation rate, while applications hosted in the New York data centre are only running at a 25 per cent utilisation rate. GSLB can use this information to direct future requests to a data centre with lower utilisation rate, in this case New York.

GSLB solutions can perform health checks against application instances that are being load balanced to determine their availability. They can identify resources and applications that are suffering from an outage or that are overloaded.

Most GSLB solutions use the Domain Name System (DNS) protocol as an underling technology for directing client requests. When a client attempts to connect to an application, it needs to first determine the Internet address (IP address) of that application. The function of DNS is to translate the application name (domain name) to an IP address. As an example of this, to access the Google mail application you need to first determine the IP address associated to mail.google.com. GSLB adds health check and load balancing capabilities to DNS and ensures that the IP address provided by DNS is associated to responsive server.

GSLB solutions are offered in many forms. Traditionally, organisations deployed GSLB as a standalone application. However, in today's high-stakes environment, many organisations prefer to deploy GSLB and DNS as a consolidated dependable solution. NC