Global Server Load Balancing - How It Works

Many organizations have operations requiring their IT systems to be available 24x7. A significant subset also needs to have their IT systems spread within their local area or globally across the AMER, AMEA, or APAC geographic regions. 

In the hybrid world that most organizations now inhabit, application servers and other core server infrastructure often gets deployed via a mixture of traditional private data centers, public cloud platforms, and some private cloud infrastructure hosted by dedicated hosting companies. This deployment model allows IT teams to deliver resilience and performance across their IT systems. 

This deployment scenario provides many benefits, such as top-notch performance for users accessing applications, plus low latency on connections as users connect to the closest data center or cloud service. It also makes delivering high availability possible as server infrastructure is not all at one location. However, this flexibility and the benefits that flow from it also introduce additional complexity related to managing and coordinating data exchange between the multiple hosting locations and ensuring that users can always connect to a site that delivers them the best application experience. Global Server Load Balancing (GSLB) simplifies the management of this complexity.

GSLB can also route network traffic and user access requests between global data centers and cloud hosting locations based on user load, even when all sites are operating. This allows the GSLB engine to spread the load across the available resources and route requests to a remote data center further away if those servers are under lower pressure and can therefore respond more rapidly than a busy local resource.

What is a global load balancer? GSLB is delivered at the infrastructure level by load balancer servers with additional software installed and enabled to provide enhanced geographic load balancing functionality. That is to say, GSLB builds on and enhances the load balancing functionality that shares access requests and connections across a local server pool to share them across multiple geographically spread server pools. These pools can be in the cloud or on-premise.

GSLB on LoadMaster extends the health checking algorithms used to determine server health to perform advanced health checking at the site level. In this case, a site is any geographic location with server pools providing web and application servers for users or other systems. The health checking uses the Fully Qualified Domain Name (FQDN) for each site as the basis for traffic routing. If a site is unavailable or servers located within it are underperforming, then requests using that site's FQDN are automatically routed to another site by rewriting the FQDN in the requests. As such, GSLB is integrated with and relies on DNS functionality and will return the IP address of the server in a site that a connection should use. The end-user (or other application services) then uses the IP address to set up a connection and communicate with the web application directly. 

As with local site load balancing, GSLB can use multiple algorithms when determining how to route network traffic between sites (see ref 1). Including round robin, weighted round robin, fixed weighting, real server load, location-based, and proximity. "Round Robin" load balancing can be used for all active sites, including support for weights and chained failover options for disaster recovery. Location Based load balancing allows GSLB to direct a client to a site based on the client's country or continent, as defined in policies. Proximity takes Location Based one step further and allows for longitude and latitude granularity in the definition of proximity. LoadMaster GSLB also offers a "Real Server Load" load balancing option, in which it uses data center metrics provided by local LoadMaster instances to allow client requests to be routed to the least busy site if required. See references 2 and 3 for general and technical overviews of LoadMaster GSLB. 

Performance - GSLB across sites ensures that each one is monitored for performance issues and traffic routed to other sites if a particular one is overloaded. Note that this is site routing based on the performance once the local load balancers at that site have spread the load across the server pool in the best way possible. This routing of requests based on site performance helps deliver an optimal end-user application experience.

Scalability - This ability to spread the load across sites also provides scalability benefits. If there is a spike in access traffic on a particular site due to a local holiday or sale event, for instance, then rather than having to deploy more web servers locally to handle the traffic, any load that local servers can't handle will be routed to another site where servers are less busy. Over time this means that each site can have fewer servers deployed, as spikes in traffic to any one site can be spread across the global server pool if required. Meaning existing infrastructure is better utilized and gives a better ROI, as well as delaying or eliminating the need for additional server deployments at a site.

Resilience - GSLB makes it much easier to deliver organization resilience by providing disaster recovery failover (both active-active and active-passive) between geographically separate sites and cloud services. See ref 2 for a detailed discussion of this topic.

Regulatory Compliance - Many countries or supranational bodies like the EU have regulations on data and how it can be stored and transferred. You can configure GSLB to adhere to these regulations to ensure that users within a particular jurisdiction do not connect to servers on sites where they shouldn't store their data. For example, users in the EU should only use sites in the EU and not USA-located sites.   

The Harris County District Attorney’s Office - Based in Houston, Texas, the Harris County DA Office serves over 4 million residents. To deliver investigative and justice-based services to this population, they make extensive use of IT systems for case management, constituent services, and employee accountability. These systems have to be available 24x7. They operate two data center sites to deliver IT services — a primary site in Houston and a remote failover site managed by a hosting provider. The latter site must provide all services available from the main site in the event of any incident that triggers a failover of services. 

After evaluating options on the market, the Harris County DA Office's IT team selected LoadMaster GSLB as the best solution to deliver their requirements. Progress Professional Services team worked with the DA Office's IT team to design, select, and deploy the best fit LoadMaster solution to deliver on their needs. After completing the project successfully, Rurik Wilmot, Senior Systems Administrator for Harris County, said: Our technical liaison was very knowledgeable and personable and made themselves available at all hours when we were to make changes to our production environment and test those changes.

Read about this project and the deployed solution via ref 4.

Even if you are using public cloud services to host your business applications, you may need to plan for a cloud outage or spread your service use across vendors to avoid lock-in. GSLB allows you to spread your cloud deployments across multiple sites and ensure that users get routed to the optimal one 24x7.

In today's business landscape, for many organizations the question isn't do we need GSLB, but rather how do we deliver resilient services without it? The question then becomes, how to choose a load balancer with GSLB? Making this choice will depend on your current infrastructure deployment, future plans, budget, and the technical resources available to deploy and manage a GSLB solution.

Progress LoadMaster has all the functionality needed for both local site load balancing and GSLB, and it delivers it at an industry-leading price via our innovative licensing model (ref 5). But you don't have to take our word for it. With a score of 4.9 out of 5 from over 360 reviews for LoadMaster on Gartner Peer Insights, you can see that the market has embraced LoadMaster (ref 6).

If you have any questions about LoadMaster GSLB, you can arrange to talk to an expert via our Contact Us page. You can also try out LoadMaster and GSLB yourself via our fully functional 30-day free trial of a Virtual Load Balancer via https://sso.kemptechnologies.com/register/kemp/vlm.