	ASx HotSync
	ASx Product Information
	Home
	Search

HotSync - Mission Critical Protection

Article by Matthew Gaunt

For a software application that presents services to a financial institution's customers, high availability is not enough, it simply must not fail! Every single second that a system is unavailable means loss of revenue and a degradation of customer confidence. Whilst any organization may accept scheduled downtime for their internal software applications, those applications which expose the organization's business to the public are genuinely mission-critical.

Therefore, historically, many institutions have chosen to deploy their mission-critical applications on legacy fault-tolerant hardware -- at a cost and without true consideration to contingency in the case of catastrophic failure. Even the best fault-tolerant hardware available would not sustain an application should the data center housing it go entirely offline or be made completely unavailable. Although this is not an argument against the use of fault-tolerant hardware, as it clearly has a place in the industry, it does seek to emphasize a change in the viewpoint of the IT industry in this decade - from fault-tolerant computers to fault-tolerant business.

At EPS our engineers were faced with a challenge: How to match and improve on the reliability offered by fault-tolerant hardware systems, whilst leveraging the mass market of open, cost-effective Unix/Linux hardware platforms.

It was clear that leaving the concerns of reliability to the hardware suppliers was impossible and that the software application must be responsible for managing its own reliability. Consequently it must not only be responsible for managing the business functionality, but also must manage the overall business integrity too.

The eps answer is "HotSync" -- an intelligent application solution, unique across the industry.

Four tiers of protection

ASx HotSync is an integrated attack on the four areas of potential system outage:

1. Software Failure

Hardware is not the only failure-prone component of the system, as anyone involved with applications running on fault-tolerant hardware will confirm! ASx HotSync acknowledges that even the most rigorously tested software may occasionally encounter a circumstance where it cannot continue. Environmental problems, operating system exceptions or simple human error may all result in failure. Any consolidated approach to providing overall failure tolerance that does not accept this may be disregarded. ASx protects availability across and during software component failure by:

-	Providing multiple redundant transaction pathways between computers in the ASx cluster
-	Providing a managed run state of all software components, intelligently restarting any failed components such that they may continue from a "clean state"
-	Maintaining data integrity should a component fail part way through completing a transaction
-	Monitoring individual components for multiple failures -- if it failed once, it may fail again immediately, unless the reason for failure has been removed -- and providing alternate processing which avoids the failed component

2. Hardware Failure

Under normal operation, ASx HotSync ensures that critical financial data is mirrored between nominated partner systems in the ASx cluster. Should hardware or software fail, transactions are routed to the partner system, and the data are uniquely marked and stored for later synchronization. When the partner system becomes available once more, it is re-synchronized with its partner. However, although straightforward in concept, this solution offers challenges, which require that the mechanism to be aware of the nature of the data that it is protecting. Simply "queuing the data to the backup system" is insufficient. There are many products in the market that can perform such a blind task.
Unlike these pure data replication solutions, ASx HotSync understands the business it is supporting and so can perform appropriate, financially accurate actions, even in the following cases:

-	When cancellation messages (reversals) are received for transactions which are not synchronized
-	When account update messages are received out of sync
-	When batch processing is run independently on multiple machines in the cluster

3. Disaster recovery

Hardware fault tolerance offers resilience over component failure in a computer. What it does not offer is full disaster recovery, should, for example, a DP centre be struck by fire or flooding. In this case, the typical response is to offer a backup computer in an alternative site.

-	ASx HotSync provides a 100% Hot Backup solution on a remote machine
-	No manual switchover is necessary in the event of total machine failure
-	ASx HotSync has been successfully deployed in environments providing hot data mirroring between sites over 250Km apart

4. Scheduled Maintenance

The benefits of using ASx HotSync mean that the full availability of a business service can be maintained - even when systems are restarted for software maintenance and enhancement procedures. Stopping an ASx HotSync system to install a new software version simply diverts traffic temporarily to the partner system. To the outside world, no loss of service is visible.

Summary

Users of transaction processing applications now expect more from the overall solution. It is no longer sufficient to have highly available individual components within the solution - the overall business itself is now expected to be highly available. The responsibility for reliability has now shifted to the application providers and they must respond accordingly. The eps response is HotSync, the only solution available today which continues to service an institution's customers during component failure, total machine failure, software failure or complete data center failure -- even during planned outages. ASx HotSync is the only data-hardening solution available today that incorporates intelligent awareness of the payment systems business, guaranteeing financial integrity ASx HotSync is the only solution available today that operates across the most popular products from hardware, software and relational database vendors.