National ICT Australia (NICTA) is set to privatise an internally developed technology that allows companies to analyse and monitor large, complex IT systems.
The performance modelling technology, known as Performance Assessment for Service Architectures (ePASA), allows organisations analysis system performance details and identify potential failure points.
Possible changes to an IT system are simulated rapidly so that system alterations and upgrades can be implemented safely, avoiding catastrophic meltdowns.
The technology piqued the interest of performance management firm Compuware earlier this year, which signed an agreement to implement NICTA's platform as an additional tool while providing a new data source to help improve ePASA's reliability.
According to NICTA senior researcher Dr Jonathan Gray, ePASA had become a prime candidate for spinning off as a private company or licensing to a third party by early next year, having grown to six dedicated programmers over the past five years.
Gray said ePASA’s promise was yet to peak, with increasing interest in cloud implementations only pushing forward performance modelling requirements.
While cloud operators could use the technology to monitor and reduce application latencies, cloud users could simulate application migration to a cloud platform in order to determine what trade-offs would be required.
“The key metric is the citizen’s experience – availability and response time,” Gray told iTnews.
“Quality of service is critical. Government agencies often cannot be sure how it will perform. Traditional stress testing cannot simulate what happens on a national scale or how it responds when peaks of activity may occur.”
Gray cited the Australian Tax Office's online tax platform eTax as an example of peak workloads, which coincide with the start of the new tax year on July 1, but only get bigger when standard users must submit their tax return on October 31.
“We call this the 'Tsunami of Business load',” Gray said.
ePASA can inform a CIO well in advance of a system roll-out and predict the most likely failure points and under what conditions. Organisations can scale up their systems or lower expectations accordingly.
“You could play 'what if', such as 'what if you took out the virtualisation layer?'"
"What if you scaled up the server farms? What if you enlarged the communication pipes? What if you used a different database technology or a different enterprise service bus?”