HCL Informix Replication Technologies

                                                                                                            Updated: 8/30/18
INDUSTRY TERMS
Recovery time objective(RTO), and Recovery point objective(RPO) are two important terms businesses use while building fault tolerant applications, and databases.
 
RPO: How much data are you willing to lose?
RTO: How much time to recovery from a failure?​

​CAP Theorem

CAP theorem states that in the presence of a network partition, one must choose between consistency and availability. Look here: https://en.wikipedia.org/wiki/CAP_theorem
​

Business Requirements for Data Replication

HCL Informix has built-in support for variety of business requirements related to fault tolerance, data replication, and database scale-out using sharding technology.
 
1. For transactional database, businesses demand zero data loss, and less than one minute recovery time in case of failure

2.  Businesses demand for Rolling application upgrade, and rolling server upgrade – No planned downtime for maintenance operations

3. Business do require transaction databases to be protected from natural disasters, and human errors
 
4.  Businesses do require transactional databases to scale out quickly to handle spikes during special events
 
5.   For geographically distributed applications, businesses do require transactional database to be available even if network goes down between distributed applications for long period
 
6.   For regulatory reasons, businesses do require certain data to be stored in the country and replicate data to central site

Customers rely HCL Informix replication technologies to meet each one of the above business requirements.
​

Physical and Logical Replication

HCL Informix replication technologies broadly classified into physical and logical replication. With physical replication, logical log records get replicated to secondary server, and secondary server continuously apply logical log records.
 
Physical replication requires database storage, hardware architecture, operating system and HCL Informix server version to be same.
 
Logical replication does capture data from logical log records, but log record data gets converted to a format understood by SQL engine. Logical replication does support heterogeneous hardware and software. Customers can pick and choose which tables to be replicated, and can replicate only few columns or few rows in a table.
 
HCL Informix High Availability Data replication(HDR), Remote Standalone Secondary(RSS) and Shared Disk Secondary(SDS) servers fall under physical replication category.

​HCL Informix flexible grid technology fall under logical replication category.
 
HCL Informix HDR, and SDS technologies support zero data loss (RPO=0) configuration.
 
HCL Informix RSS protects transactional database from natural disasters by allowing it to be deployed in wide area network environment across cities or even across continents. RSS does protect transactional database from human errors with “delayed apply” configuration.
 
HCL Informix shared disk technology(SDS) allows to quickly scale-out transactional database during special events. Since SDS share same disk as primary server, new secondary servers can be instantiated within few seconds.
 
From CAP theorem perspective, HCL Informix physical replication category falls under CP system. In case of network partitioning, secondary servers cannot accept write activity till new primary server was elected. With physical replication, primary server is the designated write server for transactional data.
 
HCL Informix flexible grid technology supports near-zero downtime maintenance operations like rolling application and schema upgrade, rolling server upgrade, and thus eliminate requirement for planned downtime for maintenance operations.
 
Flexible grid technology does not have a concept of designated write server for transactional data. Due to this reason, in case of network partitioning, all servers accept transactional workload.
 
From CAP theorem perspective, Flexible grid falls under AP system.
 
Look at this GitHub repository for hands on lab instructions to demonstrate rolling application upgrade using HCL Informix Flexible Grid technology. https://github.com/nagaraju-inturi/informix-rolling-application-upgrade
​

Database Sharding

Apart from these replication technologies, HCL Informix does support database sharding to scale-out NOSQL, Timeseries and OLTP workloads. With HCL Informix sharding technology, businesses can store massive amount of data across multiple shard servers using commoditized hardware, and query data in parallel.
 
Using Parallel shard query functionality, business can generate reports by querying data across multiple regional servers without needing to replicate and consolidate data at a central server.

​Connection Manager

Connection manager is one of the distinguished technology in HCL Informix. With connection manager, customers can configure applications to connect to a service level agreement (SLA) defined by the business rather than to a specific host and port number of transactional database. For example, OLTP applications can be configured to connect to primary server, and reporting application can be configured to connect to secondary server. Application connectivity can be configured based on geography as well. This logical separation of application connectivity from database cluster allows transactional database to failover transparently, and support zero-downtime maintenance operations like rolling application and server upgrades.
​
 Avoiding split brain situation during automated failover: In physical replication scenario where there is a designated write server for transactional data, split brain situation is one of major nightmare for DBAs in case of network partitioning. Connection manager helps avoid this problem by monitoring activity at both primary and secondary servers, and failover primary server role only when majority servers agree that primary server isn’t reachable.
 
 Automated failover configuration with connection manager helps with reduced recovery time objective (RTO less than 1 minute), and avoid split brain situation.
​

Building Fault Tolerant Cluster in Less Than Five Minutes!

With Kubernetes, Docker containers and HCL Informix clustering technology, customers can build production ready fault tolerant cluster in less than 5 minutes! Here is a reference article on how easy it is for customers to build fault tolerant cluster using HCL Informix.
 
http://www.informixcommunity.com/blogs/deploy-fault-tolerant-informix-cluster-using-kubernetes-platform
​

HCL Informix Smart Triggers

Applications can subscribe to changes within transactional dataset using SQL statement along with where clause condition, and let database server push out committed changes to applications instead of applications polling for data.
 
This light weight subscription model allows applications to register smart triggers, and get real time changes from database server without impacting client applications making changes to transactional data.
 
Enterprise Architects like push based subscription model, and dynamic nature of smart triggers where applications can register and de-register smart triggers without having to make changes to database schema.
 
Some of the use cases for smart triggers include:

1.  Exception handling in business process
2.  Real-time data flow between components in micro services architecture
3.  Integration between transactional data and blockchain smart contracts
4.  Audit on transactional data
 
Please refer to the following blog posts for details on smart triggers:
​
http://www.informixcommunity.com/blogs/introducing-informix-smart-triggers
http://www.informixcommunity.com/blogs/push-data-to-client

HCL Informix Replication Distinguished Features

1.  Smart triggers (or post commit triggers) for applications to subscribe to a database using call-back mechanism
2. Rolling application and servers upgrades using Flexible grid and Connection manager technologies – No planned downtime!
3.  Simplicity in setting up production grade fault tolerant database cluster with zero data loss configuration 
​

HCL Informix Replication Reference Architecture

Nagaraju Inturi
​Senior Solutions Architect at HCL Informix
​Connect with me on LinkedIn​​

Informix is a trademark of IBM Corporation in at least one jurisdiction and is used under license.