Replica Set Deployment Architectures¶
The architecture of a replica set affects the set’s capacity and capability. This document provides strategies for replica set deployments and describes common architectures.
The standard replica set deployment for production system is a three-member replica set. These sets provide redundancy and fault tolerance. Avoid complexity when possible, but let your application requirements dictate the architecture.
Determine the Number of Members¶
Add members in a replica set according to these strategies.
Deploy an Odd Number of Members¶
An odd number of members ensures that the replica set is always able to elect a primary. If you have an even number of members, add an arbiter to get an odd number. Arbiters do not store a copy of the data and require fewer resources. As a result, you may run an arbiter on an application server or other shared process.
Consider Fault Tolerance¶
Fault tolerance for a replica set is the number of members that can become unavailable and still leave enough members in the set to elect a primary. In other words, it is the difference between the number of members in the set and the majority needed to elect a primary. Without a primary, a replica set cannot accept write operations. Fault tolerance is an effect of replica set size, but the relationship is not direct. See the following table:
|Number of Members.||Majority Required to Elect a New Primary.||Fault Tolerance.|
Adding a member to the replica set does not always increase the fault tolerance. However, in these cases, additional members can provide support for dedicated functions, such as backups or reporting.
Load Balance on Read-Heavy Deployments¶
In a deployment with very high read traffic, you can improve read throughput by distributing reads to secondary members. As your deployment grows, add or move members to alternate data centers to improve redundancy and availability.
Always ensure that the main facility is able to elect a primary.
Add Capacity Ahead of Demand¶
The existing members of a replica set must have spare capacity to support adding a new member. Always add new members before the current demand saturates the capacity of the set.
Determine the Distribution of Members¶
Distribute Members Geographically¶
To protect your data if your main data center fails, keep at least one member in an alternate data center. Set these members’ priority to 0 to prevent them from becoming primary.
Keep a Majority of Members in One Location¶
When a replica set has members in multiple data centers, network partitions can prevent communication between data centers. To replicate data, members must be able to communicate to other members.
In an election, members must see each other to create a majority. To ensure that the replica set members can confirm a majority and elect a primary, keep a majority of the set’s members in one location.
Use Journaling to Protect Against Power Failures¶
Enable journaling to protect data against service interruptions. Without journaling MongoDB cannot recover data after unexpected shutdowns, including power failures and unexpected reboots.
All 64-bit versions of MongoDB after version 2.0 have journaling enabled by default.
Replica Set Naming¶
If your application connects to more than one replica set, each set should have a distinct name. Some drivers group replica set connections by replica set name.
The following documents describe common replica set deployment patterns. Other patterns are possible and effective depending on the application’s requirements. If needed, combine features of each architecture in your own deployment:
- Three Member Replica Sets
- Three-member replica sets provide the minimum recommended architecture for a replica set.
- Replica Sets with Four or More Members
- Four or more member replica sets provide greater redundancy and can support greater distribution of read operations and dedicated functionality.
- Geographically Distributed Replica Sets
- Geographically distributed sets include members in multiple locations to protect against facility-specific failures, such as power outages.