metamorworks -

APAC has one of highest variations in cloud performance

The network performance of public cloud services in the Asia-Pacific region has greater variability than elsewhere, new study finds

The network performance of public cloud services in Asia-Pacific (APAC) has one of the highest variations in the world, a study has found, suggesting the need for suppliers to shore up their connectivity infrastructure to keep up with growing demand for cloud services in the region.

Conducted by ThousandEyes, the study tracks the network performance of Amazon Web Services (AWS), Google Cloud Platform (GCP) and Microsoft Azure, along with Alibaba Cloud and IBM Cloud for the first time this year. Access to detailed ThousandEyes cloud and internet analyses starts at around $50,000 a year.

Measured over a four-week period, the average inter-availability zone latencies of the cloud services ranked as follows – GCP, 0.52ms; Microsoft Azure, 0.72ms; AWS, 0.81ms; Alibaba, 1.22ms; and IBM, 1.68ms.

The findings were derived from aggregated, de-identified telemetry data comprising more than eight billion service paths per day and more than 33 million network traces per hour, to provide a macro view of global internet outages.

Alex Henthorn-Iwane, vice-president of product marketing at ThousandEyes, said in Australian and New Zealand (ANZ), organisations – like those elsewhere – are trying to digitally transform and migrate to the cloud.

What had been missing, he said, was benchmarking that allowed enterprises to compare the network performance of different cloud services.

“[The study] gives IT leaders content and some really surprising information to help make their decisions,” he said.

Angelique Medina, ThousandEye’s director of product marketing, said the study’s findings were particularly important to ANZ businesses, given the greater service variability seen in APAC compared to other markets.

Notwithstanding the varying performance of cloud services, at least in APAC, the top two cloud suppliers – AWS and Microsoft Azure – delivered more stable services in the region over the past year.

AWS generally offered low latency, with a 42% reduction in variability compared to the previous year. However, its reliance on the internet, rather than its own backbone network, affected performance predictability, the study revealed.

Microsoft Azure was credited with a 50% improvement in performance predictability in Sydney in 2019 compared to 2018, while Google has strong performance in most regions. The network performance of Alibaba Cloud and IBM Cloud were described as comparable to that of the top three cloud suppliers.

ThousandEyes is currently undertaking more analysis to get a more granular country-by-country understanding of performance. Henthorn-Iwane said this was important for enterprises running hybrid cloud, as the performance data would support decision-making around where and with which cloud suppliers a company should go for.

According to Nutanix research, some 85% of companies would opt for hybrid cloud as their preferred architectural approach. At Australian carrier Qantas, for example, 30% of workloads still reside on-premise, with the rest running off public cloud services.

At an IBM event in Sydney, Qantas’ CIO Susan Doniz said: “We need the cloud to drive a sustainable competitive advantage for Qantas, because it allows us to scale and address the big complex problems.”

In Australia, demand for public cloud services is growing faster than the worldwide average, with $8bn earmarked for public cloud investment in 2020, according to technology analyst firm Gartner.

Read more about cloud in APAC

Read more on Network monitoring and analysis

Data Center
Data Management