Data Centers

Data center black magic: Load balancing and its impact on your business

Business leaders shouldn't be in the dark on important tech. Here's a brief introduction to load balancing and how it affects their businesses.

serverroom.jpg
Image: iStockphoto/amoklv

Far too many business leaders' understanding of technology in the workplace starts and ends with their personal devices. But, almost every piece of technology within an organization has an impact on the business or affects the bottom line.

What happens in the data center isn't black magic, and understanding some of the processes that go on there will help you gain a better understanding of how your business works and how to better serve your customers.

Especially, it pays to gain a clearer picture of how the data center takes in, and directs, requests from your site. And, a big part of that process is load balancing.

Basically, load balancing does just what its name implies — it is a process by which workloads, typically requests from your website, are balanced and distributed across multiple servers.

"It's really about just getting the most efficient path for a workload to go through the assets that you have," said Jeff Klaus, general manager of data center solutions at Intel.

At its most general point, load balancing can be split into two categories — local load balancing and geographic load balancing. Let's start with local load balancing.

Local load balancing is load balancing within a single datacenter, and it is what most people are referring to when they talk about using load balancing. It works like this — requests come in from the website and hit the load balancer, which then distributes the requests among a pool of backend servers to fulfill those requests.

A subset of load distribution, load balancing uses logic to determine where incoming traffic is distributed, said Robert Lacy, enterprise architect at GoDaddy. So, for example, you can tell a load balancer to automatically default to its fastest responding server, or to send traffic to the server with the least work going on. There are many other ways to configure your load balancers, but that's for another article.

Bret Palsson, manager of site reliability at LinkedIn, said an organization is ready for load balancing when it starts to notice certain red flags that could indicate a bottleneck somewhere in the infrastructure. These warning signs could include an application taking too long to load or slow results from looking up a customer in a database.

Load balancing can be implemented in a variety of ways such as software on server, through dedicated hardware, through a virtualized system, or through a system such as AWS elastic, which some would put in its own category.

One thing's certain, though. For many data center veterans, load balancing is virtually inevitable.

"There's basically never a situation that you want to have a service that is provided by single backed only and doesn't have a load balancing layer in front of it," said Nic Benders, director of engineering at New Relic.

There are a variety of reasons that load balancing is important and they all affect the business in one way or another. The two main reasons are availability and growth.

"Availability ensures that, in the event of failure, systems continue to function as expected," Lacy said. "While many architectural patterns exist to address availability, load-balancers can be configured to automatically stop sending traffic to unhealthy back end systems until the problem is addressed."

What this means is that, with proper load balancing logic, your website won't crash or you won't lose specific features if a particular server fails, as the load balancer will redirect traffic to the remaining backend systems. So, it limits your points of failure and can potentially give you better uptime on your site.

However, that puts a lot of pressure of the load balancing tier, so it's got to be more resilient than the systems it is sitting in front of, Benders said.

In addition to protecting your product, load balancing can also contribute to growth and scalability. Almost every organization will get to the point, typically pretty early on, where they can no longer serve users with a single machine.

Having load balancing in place means you can more easily add servers to your organization, a process known as horizontal scalability. This is different than vertical scalability, which is replacing existing servers with more powerful ones.

It also allows you to use them more efficiently.

"Most of the literature in the data center will tell that CFO or the [operations] person that most data centers have a good 10-15% of underutilized capacity," Klaus said.

Using load balancing, you can repurpose that capacity and add it into the balancing act. This directly affects your business's bottom line.

It's important to the frontend user experience too.

"It gives you the ability to change the way that your service is implemented in the backend, without exposing any changes to the people who are consuming your service," Benders said.

Load balancing allows you to change backends during deploys and change the way traffic works with filters and rules and a plethora of other capabilities.

The other type of load balancing is geographic load balancing.

Benders said geographic load balancing is "Where you have multiple whole, isolated datacenters and you're taking your incoming traffic and directing either based on the fasted geography or based on an even load out to multiple datacenters."

This type of load balancing is much more complicated and many organizations and businesses never even get to the point where they have to consider it. Still, it's important to have a basic understanding of how it works.

For starters, Bender said, you can run multiple live data centers or live and offline. A lot of times, this is done for security and efficiency purposes.

If you have multiple active data centers, you will commonly see more than two, because if you are balancing workloads across two data centers and one fails, it may be difficult for the remaining one to handle the load. However, if you have three or more data centers and one fails, it's easier to distribute the work across the remaining ones.

Benders said that another reason for multiple datacenters would be to handle a jurisdictional restriction on your company.

For example, hypothetically, privacy laws may dictate that European customer data must remain in Europe, meaning you would need a data center there to house that information.

Performance also comes into play relative to geographic load balancing. An example of this would be having a data center in Asia to better serve your customers in Asia with lower latency.

Hopefully this has given you a broad overview of load balancing and why it is an important concept to grasp in your business.

Also see

About

Conner Forrest is Enterprise Editor for TechRepublic. He covers startups and enterprise technology and is passionate about the convergence of tech and culture.

Editor's Picks