In-Memory Data Management (IMDM) and high-availability sound like an oxymoron but with a need for low latency in several mission-critical applications, high availability has become a mandatory requirement. The advances in IMDM technology and the evolution of IMDM architectures have made high-availability possible even though data is stored in-memory. This can be achieved in several different ways:
- Data Persistence – Data maintained in-memory is also persisted on disk every few milliseconds to avoid any data loss in case of a server crash. The server is designed to auto-recover by loading up the persisted data on disk.
- Data Redundancy – Data is replicated across multiple servers to avoid any impact to the application if one of the servers goes down.
- Fail-Over or Mirror Servers – Each server is configured to have at least one fail-over or mirror server which takes over the application requests in case the active server goes down. This automated transition does not impact the end users.
- Disaster recovery setup – The complete IMDM setup can be replicated across data centers to support high availability in case of a data center outage. This can be an active-active data center configuration to distribute load or can be used only in case of a disaster.
BigMemory supports all the above ways for high-availability and has been deployed at several locations where achieving five 9′s of uptime is a key requirement. If your environment is challenged with any performance issues and you are concerned about availability before deploying an in-memory solution, please contact us to discuss.