Multi AZ Deployment

  • Thursday, Sep 12, 2019

Why More Cloud Is Always Better

Taking Advantage Of Redundancy And High Availability

The Cloud has revolutionised the way we use and interact with IT systems. It’s no longer about capacity planning for that big project coming 12 months from now, or needing to have staff on call for that Black Friday event to help manage the thousands of customers taking advantage of sale discounts.

It is now easier than ever to get what you want, when you want it, and most importantly, for how long you need it, then — keeping on with sales terminology — return it to the store when you no longer want it.

But even with all the possibilities and all the new levers and pulleys made available to enterprises and weekend warriors alike, not many are taking advantage of them and some don’t even know they exist.

I will be covering some of the advantages that the Cloud can offer you and walking you through them.

How is the cloud different to normal hosting?

Originally, when someone wanted to implement an IT service, they needed a physical place to host the server. This place ranged from a spot in their garage to an entire data centre.

Having everything within arm’s reach sounds great, but what happens when that data centre or home-hosted server fails. Now your IT service no longer provides that value. It’s costing money with no return.

So where do we go from here?

In the cases of enterprises they would go and buy two data centres for redundancy. For people who have their server running in their garage, some would wear the loss in value while others would look to host their service with a business — usually one that provided the high availability and redundancy they did not have originally.

Obviously, with doing this there comes a massive cost hit, so the value of your IT service is now reduced because of the added outgoings. If this is starting to sound familiar, you might be wondering about how the Cloud could possibly help. Isn’t it the same thing?

Let’s look at Amazon Web Services (AWS) in Australia.

At the time of writing, there are three availability zones Amazon has provisioned for the country. Availability zones are data centres that are networked together to provide computation, storage, and redundancy services beyond what traditional hosts could offer. But, throughout my time as a consultant and working in IT (which would be close to 10 years), I have come across a worrying number of clients that would only use 1 or 2 of these availability zones.

This has always been one of the first items I engage them on.

Why should I use all three availability zones?

It used to be expensive and a lot of work to maintain one data centre, let alone two, but when anyone moves to the Cloud, in most cases they will have access to three data centres that are already networked and maintained, essentially for free for you.

Why would you not use them all?

By doing so, you have increased your resilience and availability for your IT services.

This might seem counter intuitive. You may be thinking “But running three sets of my service will also increase my cost and reduce the return of my application further”.

While you are right in thinking this, this is where we pull on one of the other leavers of cloud.

“Pay for what you need, when you need it” — in other words compute-on-demand. In having three availability zones set up to run workloads, there is a cost, but the cost is minimal at best, since having compute-on-demand means you don’t need to be running your service in all three zones, you just need to have it available so that when a failure occurs, you simply fail over to another zone.

By doing this, your operations will remain highly available and you still have another two availability zones that can handle your traffic, which means you can handle yet another failure.

So, establishing your service on all three zones means the likelihood of you having a full outage is greatly diminished.

What happens to my staging and development servers in a failure?

The other reality I point out to clients who haven’t provisioned their service on all three zones is that everyone in that zone will fail over at the same time.

Both remaining zones will be hit with a surge of traffic, and while AWS can ensure your staging and production servers are separated, if they’re only in one zone, a lop-sided traffic surge can wipe you out entirely.

You may as well have been hosting your service in your garage.

Though there are more options for increasing availability and resilience, like going multi-region, I wanted to keep this blog simple and easy for anyone that had not already taken advantage of such avenues.

(Sadly, multi-region roll-outs can lead to several complexities either technological or even legal.)

I hope this post has provided you with enough reason to consider setting up your service on all three availability zones, if not now, in the future. I also hope you have all enjoyed reading this as much as I have enjoyed writing it.