Impact of latency on database replication
Databases store valuable data, such as bank account balances, orders in a web shop, or other business-critical customer data. It is essential that databases managing this data perform optimally, have low response times, provide the most up-to-date data, and remain consistent. A commonly used best practice to achieve these characteristics is to replicate data across multiple nodes and make it accessible from geographically separate locations.
Database replication in distributed systems
Database replication is a mechanism within distributed systems that contributes to fault tolerance and high availability. As organizations operate globally, multi-availability zones are used to deploy infrastructure in different regions via cloud platforms. Database replication helps to keep data consistent between these different availability zones.
Network latency in the replication process
Because data is replicated across regions that are geographically distant from each other, the distance between these zones inevitably introduces network latency into the replication process. In high-speed networks, latency is the most dominant component of the total round-trip time (RTT). Depending on the chosen consistency model and replication setup, latency can affect database performance, data availability, and data consistency. Latency thus introduces additional complexity in the form of trade-offs between different replication and consistency models within an implementation.
Research objective
The purpose of this research was to gain insight into the impact of network latency on transactional database replication within simulated, geographically separated database environments.
