DataFlex WebApp SPLF Server Ensures 24/7 Availability

A large financial services institution in the Australasian region has significantly invested in DataFlex technologies to develop Web Service access to their core, line-of-business applications. These DataFlex WebApp web services are consumed by a constantly growing number of applications (currently about 20) via their enterprise service bus that offers 24/7 support to their internal users and customers. Some web service calls are triggered automatically (service-to-service) while others originate from users in their offices around the world (Australia, India, South Africa),

Given the size of the company and the nature of its business, downtime of the web service-based applications must be strictly limited to unavoidable situations. The company has adopted a continuous delivery discipline resulting in about 2 software releases each week and has established that deploying new releases is NOT an acceptable reason for downtime. Real and perceived high availability is their goal and the company decided to make appropriate investments to fulfill it.

After researching and implementing several server load balancing and fail-over solutions to avoid service interruptions, it was established that the DataFlex WebApp SPLF Server is not only by far their most cost effective load balancing, fail-over and release management solution, it is also the most fit for purpose. Load balancing software often requires experts to setup, monitor and keep it working properly. The power in the DataFlex SPLF Server lies in its effectiveness and its simplicity.

Here’s how the DataFlex SPLF Server is used to support continous software delivery and avoid downtime that would affect application users...

The company’s environment consists of a DataFlex WebApp SPLF "master" server and multiple SPLF "slave" servers that actually process the web services. Operationally, weekly software updates are performed using a release management tool: Octopus Deploy. Web application developers are responsible for packaging the release and, using Octopus, deploying it into the pre-production environments. The production control team is then responsible for triggering the release into production using the release script loaded in the Octopus software.

To perform the update, the release script would normally deactivate the web services application to be updated which, without special accomodation, would result in downtime. Using the DataFlex SPLF Server though, downtime is completely avoided using the following scenario:

1. The first WebApp SPLF slave-node to be updated is stopped and flushes its processes on the SPLF master server which senses the deactivated slave and then only sends application requests to other active slave-nodes in the pool.
2. While the other slave nodes manage the web service load, the deactivated WebApp slave-node is updated and, once that’s completed, it is reactivated.
3. The WebApp SPLF master is then triggered to recheck its available slave-nodes and add a number of processes from the now-updated WebApp slave node are added back into the pool.
4. Then the the process is repeated - the next slave-node is deactivated, updated and reactivated until all SPLF slaves’ software is updated ensuring that there is always at least one slave-node active and providing 24/7 availability even during the regular software updates as well as other server maintenance or unanticipated system failures.

Needless to say the customer is extremely happy with their combination of DataFlex WebApp web services applications and the SPLF Server solution that meets their software update and downtime avoidance goals with added security, performance, load balancing and general purpose fail-over benefits as part of the bargain.