Amazon Web Services goes live in Sydney

Amazon Web Services has confirmed the industry's worst-kept secret, setting live two Australian availability zones in Sydney overnight, one of them being Equinix SYD3.

Linked gallery: Photos: Equinix opens phase two of SY3 data centre

Asia Pacific (Sydney) is the ninth region in which Amazon has deployed its cloud platform.

The firm already counts 10,000 customers based in Australia and New Zealand, a substantial proportion of which are expected to move their workloads to the Sydney facility to gain better local performance or reduce regulatory risk.

Launch customers included the Commonwealth Bank, Harvey Norman and the REA Group.

The company said the AWS region will offer "single-digit millisecond latency to end users in Sydney".

Amazon vice president Adam Selipsky, speaking to iTnews from Seattle, said he expects “broad and rapid” adoption of services from the new availability zones, but did not wish to set any benchmarks or goals.

“Many customers have told us they would like us to have an AWS Infrastructure Region in Australia,” he said. “We expect a good number of [the 10,000 existing customers] to spin up new applications in the Sydney region or move applications to Sydney.”

Pricing

AWS pricing (see analysis below) for the two Sydney availability zones is roughly in line with Singapore - both of which are a 30.76 percent premium on using AWS zones based in Northern Virginia and Oregon in the United States.

“We have a very consistent pricing philosophy,” said Amazon vice president Adam Selipsky, speaking to iTnews from Seattle.

“When we price, we look at power and transit costs region-by-region,” he said. “Regions with higher tax structures or higher transit costs tend to be priced higher.”

A default "small" on-demand instance costs US$47.48 a month in Northern Virginia or Oregon, or US$62.09 in Sydney or Singapore — a 30.76 percent difference.

The 30 percent premium remains for monthly price comparisons for standard medium, large, and extra-large instances. Those instances in Sydney cost US$124.19, US$248.37 and US$496.74 a month respectively.

Selipsky said part of the company’s success owes to its ability to use scale to lower costs from its hosting and telco suppliers.

“We have strong relationships with bandwidth providers in each of our regions. We use our scale to drive down our costs by negotiating and managing these relationships. We engineer our internal costs as low as possible, and instead of pocketing the margins, we lower prices,” he said.

Amazon Web Services is on a run of 23 successive price reductions in the United States.

Selipsky acknowledged that many telco providers are also now in the business of providing competing cloud services. “We co-operate in some areas, we compete in others,” he said.

Selipsky said Amazon is transparent about offering volume discounts to enterprise customers. “In general, an organisation’s costs gets lower as their use increases. We offer volume discounts across S3 storage, bandwidth and data transfer charges and volumes of reserved instances.

“Beyond that, absolutely, we can also talk privately with large organisations about price.”

Service availability

Not all services available in Amazon's US-EAST availability zones are being made available in the Sydney facilities.

For example, high-level processing instance types available in North Virginia, such as second-generation (M3) standard instances, cluster compute, cluster GPU and high I/O instances are not available from Australian data centres.

Support

Amazon has offices in three Australian cities and plans to open a local technical support operation to launch in Australia early next year, which will form part of a global network of support centres.