How to Manage Fault Tolerance with Availability Zone and Regions in AWS

What are Regions and Availability Zones in AWS

AWS infrastructure is available across various geographical region.  Usually regions are across continents of countries but for large countries like US there can be multiple regions.  Each regions is divided into 2 or more Availability zones which are nothing but collection of data centers which are connected to each other with high speed connections.

These data centers are separate units which are located away from each other so that if any calamity happens around one of them the other data centers can take care of the load.   These data centers are located in highly secure buildings with power backups as well as multiple network connectivity.

How Regions and AZs Help in Achieving Fault Tolerance

Availability zones constains data center which are insulated from failures in other data centers.
 If a particular data center goes down then also application will not go down as other data centers are insulated.  AZs have low latency high speed connection between the data centers.

Usually fault tolerant systems have instances across different AZs in a regions.  It is highly unlikely that all the AZs in a region would go down at the same time.

Example architecture of multi AZ hosted Application

In a multi AZ hosted application where you have web, backed and database layers and the layers are hosted across two AZs in a active - active scenario then traffic is distributed by the ELB across the two AZs.  When any one AZ is goes down in that case other AZ can take over.

Another scenario is active - standby scenario where one set up is active but the other set up is is a standby mode. However data replication happens from the primary to the backup.  If active AZ is impaired then the standby database is elevated to be primary and the traffic is routed to the new database.

For active - active setup, ELB can be used to direct the traffic to different stacks which can be hosted as separate stack in same AZ or across multiple AZs.

What if You want to Achieve fault tolerance across Regions

ELB can work only within same region. For fault tolerance across multiple regions you would need Route 53.

Comments

Post a Comment

Popular posts from this blog

Some Important Architectural Considerations

As a Cloud solution architect when considering cloud migration here are some important considerations and questions to ask - Disaster recovery For how long can the application be down?   Do you need the application to be up 24/7 no matter what the cost is?   What are the possible locations from which client is willing to operate.   Is it only a single state/single country or having locations across the oceans or continents. For backup sites do you want it to be active/active in which case application can be recovered in matter of minutes by switching to another site.  Or do they want to keep snapshot of database backups in which case it may need more time like few hours to recover. Business Continuity Can the business be offline for few minutes/hours/few days in a week or Customers use it 24/7? Cost How sensitive if customer towards Cost.  Are they ready to compromise of Disaster recovery time or is it other way round. Is there a...
Read more

Role of AMIs and EBS in designing Fault Tolerant Systems

What is Fault Tolerance? Simply put it is the ability of the system to be in operational or be available even if at least one or more of its constituent systems fail.  So what happens if a system is not fault tolerant.  Well, the price of being down depends on the business.  In some cases it may cost few millions for few minutes of downtime. How AWS can help prevent or reduce Downtime? AWS has tools and techniques to build fault tolerant systems which can be automatic or requires minimum human intervention.  There are many such tools in AWS but in this article let us look at the role of AMI and EBS. What is AMI? AMI is Amazon  Machine Image which is basically a prebuilt software configuration which can be applied to EC2 instance.  It can include Operating system, Application servers and application apart from security patches. As an enterprise it is important for you to create your own library of AMIs which can then be used to instantiate EC2 ins...
Read more