The dangers of not having a high availability setup for your load balancer  — as well as your applications!

The dangers of not having a high availability setup for your load balancer — as well as your applications!

High Availability Published on 3 mins Last updated

Downtime sucks. Regardless of the cause. Surely the whole point of buying a load balancer is to help you avoid this? So why the heck do you need TWO of them?

Step One: Avoid a single point of failure for your applications  

Want to avoid downtime? Want to be able to serve much more content than a single web server can provide? Need the Exchange cluster to grow? Need your storage farm to provide more capacity? Then yes, sure, you need a load balancer.

Adding a load balancer removes the single point of failure from your applications. So job done, right? Well, not quite.

By adding a single load balancer, in essence, all you've done is SHIFT the single point of failure from your applications to the load balancer! So you've effectively kicked the problem down the road. No big deal if you're not worried about downtime, but isn't the whole point of load balancing to avoid this?! So why fork out on a high availability solution, if that solution isn't watertight?

What you need is a second load balancer...

Step Two: Avoid a single point of failure for your load balancer!

There are many reasons why it might be foolhardy to rely on a single load balancer (hardware failure, network failure, user error, etc, the list goes on...). The advantage of TWO load balancers is that in active-passive mode, if the active load balancer becomes unavailable, then the service is immediately pulled across to the other machine - without users noticing any break in communications at all. This is what we call 'failover'. So, while we go out of our way to protect against most issues affecting a single load balancer, without a second appliance, there's potentially a disaster (or at the very least an unforeseen event) waiting to happen.

The ability of a pair of load balancers to 'failover' from one to the other, really is necessary to protect the organization against any single point of failure. After all, the last thing we want is to introduce any single weak spot in an otherwise robust and resilient environment, right?

The failover advantage

The failover between the two load balancers is built-in and integrated and has been well tested over many years! We have alerting and monitoring on the system to let you know if there has been a problem, as well as health checks that check and verify the status of all Real Servers. No external system is required to do this.

Having a second load balancer brings other advantages too...

The maintenance advantage

Should you need to maintain the infrastructure which houses the load balancer, then having the second unit means that shutting down one at a time is not a problem. So you always have one load balancer online.

The update advantage

We also provide regular updates to the load balancer software which you can much more easily implement with two load balancers. You will certainly want to take advantage of these updates as they contain a mix of new or improved features, potential bug fixes (yes, we do occasionally have bugs, which we fix!), and security updates. And having two load balancers means that fixes and updates can be applied to the load balancer not in active use - therefore avoiding any disruption to service.

Having the second load balancer also means there's a backout plan. So should the service not work as expected for any reason, then it can be moved back to the untouched load balancer ('node'), as easily as this...

...without any difficulty from the front page of the appliance web interface.

The perfect pair

So there you have it. Two load balancers really do give you the high availability solution you're probably after. So as tempting as it might be to want to spend less and buy just one load balancer, in the end, you may end up paying much more.

Of course, our Support team are always here to advise and assist. And if you need help with any load balancer updates, we're always on hand if you want to sense-check anything.

How to avoid downtime

Removing the single point of failure