Redundancy is one of the things I had a challenging time grasping when I started exploring the cloud. The concept of redundancy is pretty simple: your data is important – you don’t want to lose important things or face downtime – so let’s have several sets of the same data exist in multiple places. The cloud offers several redundancy options that, if implemented correctly, allow full backups of your data ready to be restored quickly in case the worst happens.
We’re therefore going to start a new series where we will slowly go through the Azure Storage Redundancy options to hopefully demystify all the confusing and repetitive terms. This should help us understand better how the cloud can handle redundancy and secure our data.
So let’s try to simplify and explore the options that a public cloud service such as Microsoft Azure provides to make this happen.
Before we start it’s important to remember that the cloud is simply using someone else’s infrastructure. When we use Azure, we are using Microsoft’s cloud infrastructure. Microsoft stores their infrastructure in Data Centers which are big warehouses full of computers, networking, and storage options we can ‘rent’ as we use the cloud. It all starts with these data centers – because these are the core units of the cloud.
Starting at the top: Azure Regions
Azure Regions are, as the name implies, regions across the world where Azure has several data centers ‘close’ together. Close is defined more as a standard for low-latency connections between them, meaning that they can (and have to) talk to one another extremely quickly on the same network to be in the same region – this, therefore, means that they are also physically relatively close.
There are several Azure regions around the world such as Denmark East, France Central, or Italy North. It is through these regions, that availability of Microsoft’s Azure services is introduced to a certain area in the world. Some parts of the world have several regions such as East US, East US 2, and East US 3.
Different regions have different costs to make use of their infrastructure. When you create resources, like a Virtual Machine, in a particular region. You can choose to use the region closest to your end-users and allow your applications to have several instances that can be accessed (higher availability), but it might also cost more to host your resources there. So there’s a cost consideration that needs to be weighed against the application’s requirements.
Within an Azure region, like France Central, there are several availability zones. This means that there are 3 or more groups of data centers within the France Central region. For instance, you have:
- France Central Availability Zone 1
- France Central Availability Zone 2
- France Central Availability Zone 3
Each of these availability zones uses its own power, cooling, and networking setups so they are completely separated from each other. This is good because they are more ‘independent’ this way and less likely to be affected by similar issues simultaneously.
Now that we know what Azure Regions and Azure Availability Zones inside these regions are, we can look at two ways we can have data redundancy within a single region.
Locally redundant storage (LRS)
With a storage account available to you, by default, you will have 3 replicated versions in your data in the primary region of your choice. With LRS, this is all done within the same data center – 3 local copies. Doing it this way is the cheapest. The data centers are already like mini cities, and having 3 copies within various regions in the data center is pretty good.
Zone redundant storage (ZRS)
This is where the Azure availability zones come in. Instead of keeping the 3 copies within the single data center, we can make it redundant across the 3 availability zones within the region. This is of course a more secure option as well as useful for applications that need to be highly available, and it’s therefore also more expensive.
Pairing the Azure regions
Microsoft Azure decided that it would be a great idea to pair up the regions. This allows for some primary data recovery and business continuity in case of a data loss disaster across regions. Most of these pairs are located within the same geographical region. Pairs have direct connections but are at least 300 miles (approx 482 km). In case of natural disasters or power outages, both regions won’t go down at the same time. So when you start creating resources in the primary region of your choice, this data is replicated within the secondary region by default.
Save the Data!
Well if you’ve made it this far, I’m sure you will appreciate the rest of this series! I hope that it will help you set up your environment in a way that is more convenient for your users, secure, and cost-effective!
Next week we will focus on redundancy in secondary regions.