Cosmos DB is a buzz word in distributed system world because of its multi model, multi-master, low latency and high throughput. You can create an azure Cosmos DB account using the Azure portal, PowerShell, CLI or Azure resource manager template.              

In this article, I am going to explain about multi-master, how it solves the problem in data replication scenarios with multi-master. Before reading this article, please read my previous article to understand  basics of Cosmos DB.


In distributed system model, data availability is a big challenge with low latency. Cosmos DB already provides a solution to latency in data read with help of multi region replication and choosing opted consistency model. So, there is no latency in Read operation, because the data is available in your near region data server.  

But there is a quite bit latency in write operation because only one datacentre is there to perform write operation. Please read this blog to understand how global distribution is working.  Now Cosmos DB releases a new feature to improve write latency with help of Multi-master.  

This article will explain the nice extent of Cosmos DB’s Multi-master support which is a new release in Azure with its advantages and how it handles data replication conflict.

What is Multi-master? 

Till now all the distributed system architecture works like one write data server and multiple read data servers. Data are replicated to all data servers from write data server. 

With multi-master feature we can choose all data server act as write server data will be replicated with other data servers. The big challenge of multi-master, avoid the data duplication in data replication. 

 How to enable Multi-master features using CLI: 

Below is the command used to enable multi-master feature using CLI, 

az cosmosdb create  

   — name “myCosmosDbAccount”  

   — resource-group “myResourceGroup”  

   — default-consistency-level “Session”  

   — enable-automatic-failover “true”  

   — locations “EastUS=0” “WestUS=1”  

   — enable-multiple-write-locations true 

Let’s see in detail about each keyword in the above command 

az cosmosdb create  

azIt denotes Azure 

cosmosdb: What resource want to use in Azure 

createHere, we mention want to create the resource 

— name myCosmosDbAccount  

nameIt is a property to create cosmos DB account, should be unique. 

— resource-group myResourceGroup  

resource-group: Here mention already created resource group. 

— default-consistency-level “Session”  

default-consistency-level Cosmos DB offer five different type of consistency, we can set any one of them as default. 

— enable-automatic-failover “true”  

enable-automatic-failover: Using this property one of the read datacentre act as write datacentre 

— locations “EastUS=0” “WestUS=1”  

Locations: Here, we have to mention location to replicate the data. 

— enable-multiple-write-locations true  

Enable-multiple-write-locations: This property is main property to enable multiple write data centers. If it is false, 0th location data center will be write remaining are read. 

SharePoint Intranet Software

How to enable COSMOS DB Multi-master features using Azure portal: 

 In a new browser window, sign in to the Azure portal. 
Click Create a resource>Databases>Azure Cosmos DB. 
 Fill all the details, refer below image.   
 COSMOS DB Account name should be unique. 
Need to enable Multi-region writes, by default it is disabled. 

nable COSMOS DB Multi-master features

Replication difficulties: 

Let me explain the difficulties at replication in multi-master architecture.  

 Let’s assume we have three data server called A, B and C, all are having read and write rights. 

Data server

 Data a, b and c wrote on A, data d, e and f wrote on B, data g, h and j (here do not use i) wrote on C.  
 First, server B replicated from A by data d, e and f. After replication B have a, b, c, d, e and f.

Datecenter Second, server C replicated from B by data a, b, c, d, e and f. After replication data will be a, b, c, d, e, f, g, h and j. 

Date center replication Third, server C replicated from A, data are a, b and c. But these data already available in C server. If replication run, data will be duplicate see the below image. 

Datacenter replication

Conflict Scenarios: 

There are three types of duplication happen while working with multi-master. 

InsertIt will occur when we insert two or more documents with the same unique (for example ID) index from two or more regions simultaneously. Documents get removed when we apply conflict resolution only one document with the original ID is committed. 

ReplaceIt will occur when we update single document simultaneously from two or more regions. 

DeleteIt will occur when we delete a document from one region and update it from one or more regions. 

Cosmos DB offer three different type of conflicts resolutions like Last-Writer-Win (LWW)Custom- user defined procedure and Custom – Asynchronous  

As of now only LWW available for all the model of cosmos DB and it is default. We can use all the three only in SqlServer.  


It will resolve the conflicts based on the numeric value passed in a property on the document. 

Insert and Update conflicts: 

If one or more document duplicates in insert and replace operations, the document that contains the largest values for the conflict resolution path becomes the winner. If multiple document has same numeric value for resolution path, the selected winner is non-deterministic. All region replicated by winner region document.

Delete conflict: 

If there are delete conflicts involved, the deleted always win over other replace conflicts.

Some use cases: