web hosting

Advanced Web Hosting Technology

Along with the web hosting industry comes a lot of unfamiliar terms and technical jargon.  While some are self explanatory and a bit easier to process, others are downright confusing, especially to the newcomer who knows little to nothing about the industry.  This article will introduce you to some of the more advanced terms and explain their meanings.

Availability

Often referred to as uptime, the availability of a system describes its accessibility and ability to provide a level of redundancy that works to prevent failure at any single point.  For instance, if only one server is responding to HTTP requests, the server qualifies as a single point of failure.  If there are two web servers in a load balancing scheme, then the load balancer itself is the single point of failure.  Most high availability systems are equipped with efficient fault-tolerance mechanisms to ensure the necessary redundancies.  Because 100% availability is costly and rather difficult to achieve, the industry standard in the web hosting business is at around 99.9%.

Load Balancing

Load balancing is a method web hosting providers employ to more evenly distribute the system load over a group of available servers.  This technology is used to ensure that all servers process their fair share of requests without being overworked and more susceptible to failing.  There are many types of load balancers and these mechanisms come in both software and hardware varieties.

Response time and Transaction Load

The performance of a server is associated with the response time of a given transaction load.  In regard to a website, it is measured by the time it takes to load web pages and process requests to the web server.  The performance of large, high traffic websites can be greatly enhanced by upgrading hardware, residing on a dedicated server or utilizing load balancing technology.  In many cases, fine tuning software packages such as MySQL and PHP can also increase response times.

Scalability

The scalability of a system generally refers to the property that enables it to service any given load by upgrading hardware or incorporating additional services in a predetermined manner.  It essentially describes an instance in which performance and throughput can still be maintained once the load increases.  Scalability is a great quality to have in a web hosting solution as it can support the growth and increasing needs of your website.

Session State

Session state is often needed to maintain the connection of a client program and a given server.  This typically needs to be done to accommodate a user who is logged into the system.  If the topology consists of multiple servers, maintaining session state is crucial and something that must be taken care.  Most load balancing software applications on the market provide facilities that enable this to done.  The most common methods are implementing cookies and rewriting URLs with the server details included.  load_balance.htm

Throughput

Though closely tied to performance, throughput is best described as the number of concurrent transactions the server is able to process.  The higher the throughput, the more processing power to the system.

How to Select the Approprite Load Balancing Technology

Load balancing is a hot topic in the web hosting industry as the technology is being increasingly used in server farms and shared hosting environments.  Its purpose is to fine tune computer systems and networks in a manner that better evenly distributes data and processing across available resources to prevent a single device from being overloaded.  Some of the busier websites utilize two or more servers in a load balancing scheme. Therefore, if one is overwhelmed, traffic is forwarded to another with more capacity.

Common Types of Load Balancing

The process of load balancing is one that can be achieved in a variety of ways.  If this is a technology you are interested in, it is best to know what methods are available to you.   After learning a little more about the types of load balancing, you will should be able to better determine which option works best.

Round Robin DNS Load Balancing

One of the earliest adopted techniques, round robin DNS load balancing uses the round robin feature built into the BIND DNS server to facilitate cycles through the IP addresses corresponding to a farm of servers  This is one of the easiest and most inexpensive methods to implement.  The downside to this type of load balancing is that the DNS server doesn’t have any knowledge of the server’s availability.  Therefore, there is a possibility that data could be continuously forwarded to an unavailable server.  There is also a chance that an IP address will be cached by other nameservers and not sent to the server designated for load balancing.

Hardware Load Balancing

With hardware load balancing, TCP/IP packets are routed by a device to numerous servers in a cluster, a method that leverages circuit level gateway to route traffic.  This method is said to create a powerful topology with a high level of availability.  The one big drawback is a cost that is often considerably higher than other methods.

Software-based Load Balancing

The software-based variation is the most commonly used load balancing method.  These type of load balancers are typically integrated as components in proprietary web servers and application server platforms.  This technique is more configurable and offers the ability to implement intelligent routing based on various input parameters.  In many cases, software-based load balancing requires additional hardware be provided to isolate the load balancers.

Narrowing Your Options

The decision of choosing a load balancing configuration based on hardware or software should be made on one’s particular needs.  For instance, if your website only requires load balancing to serve HTML pages or lightweight dynamic, data-driven pages, the round robin method should be sufficient.  On the other hand, if your traffic and data processes are more demanding, you would probably get the best performance out of the more advanced software or hardware methods.  Whatever method you take, keep in mind that the load balancer should be able to deliver intelligent monitoring to evenly distribute the load.  Depending on your situation, different techniques can better direct the loads to the servers that are more capable of handling them than others in the cluster.