Developer infrastructure in Australia

Australian developers face three constraints global comparisons ignore: 150-200ms of latency to US regions, data residency obligations under the Privacy Act, and USD billing that adds an exchange spread to every invoice. Sydney regions, AUD-billing local vendors and edge networks with Australian points of presence are what actually fix them.

Editorial judgement last reviewed

Australian comparisons

What should Australian developers know about choosing infrastructure?

Most infrastructure comparisons are written from San Francisco, and it shows. They assume a US region is nearby, that USD billing is invisible, and that "data residency" is a European problem. None of those hold here.

Sydney is roughly 12,000 kilometres from Virginia. That's about 200 milliseconds of round-trip latency on a good day, before your application does anything. Singapore is closer at around 90 milliseconds, which is why so many vendors serve Australia from there and call it regional coverage. An actual Sydney presence is under 10 milliseconds from most of the east coast, and the difference is the difference between an app that feels local and one that doesn't.

Then there's the money. Almost every developer tool bills in USD, which means your infrastructure cost moves with the exchange rate and you absorb a conversion spread on every invoice. For a small company that's a meaningful and entirely invisible tax.

And there's the law. The Privacy Act and the Australian Privacy Principles govern how personal information is handled and disclosed overseas, and for government, health and financial services work, hosting location is frequently a hard contractual requirement rather than a preference.

These pages take the same comparison data as our category hubs and re-cut it for those constraints.

Latency: what the numbers actually are

Rough round-trip times from Sydney, useful for sanity-checking any architecture:

  • Sydney (ap-southeast-2 and equivalents): under 10 ms from Sydney, roughly 12-15 ms from Melbourne, 25-35 ms from Brisbane and Adelaide, and 50-60 ms from Perth, which is further from Sydney than Sydney is from New Zealand.
  • Melbourne regions: now offered by several major providers, and worth using if your users skew Victorian.
  • Singapore: 90-100 ms. Common for vendors claiming APAC coverage.
  • US west coast: 140-160 ms.
  • US east coast: 190-210 ms.
  • Europe: 250-300 ms.

The trap is that latency multiplies. A page making six sequential round trips to a US database spends more than a second waiting on the network alone, and no amount of frontend optimisation recovers it. The fix is architectural, put the database near the application, and cache aggressively at an Australian edge, rather than a matter of choosing a faster vendor.

Perth deserves a specific mention. It's roughly as far from Sydney as Sydney is from Auckland, and applications designed around east-coast latency assumptions feel noticeably worse there. If you have Western Australian users, measure from Perth rather than assuming.

Data residency, and when it's actually required

Data residency in Australia is more nuanced than "the data must stay here", and conflating the two costs money.

The Privacy Act and the Australian Privacy Principles, APP 8 in particular, govern cross-border disclosure of personal information. The obligation is largely about accountability: you remain responsible for how an overseas recipient handles the data. That's satisfiable with contractual terms and does not automatically require Australian hosting.

Where it becomes a hard requirement is sector-specific and contractual. Government work frequently mandates Australian hosting, often with certified providers. Health records carry state-level obligations on top of federal ones. Financial services customers may impose it through procurement. And plenty of enterprise contracts specify it simply because the customer's own compliance function asked for it.

The practical advice: find out whether your requirement is legal, contractual or a preference, because they have very different costs. A preference is satisfied by an edge network serving from Sydney. A contractual requirement needs the data at rest in an Australian region, documented. A government requirement may need a specific certification, which narrows your options considerably.

Note that edge caching is not residency. A CDN serving cached content from Sydney while the origin sits in Virginia gives you the latency benefit and none of the residency guarantee, and that distinction has caught out more than one procurement conversation.

AUD billing and the invisible exchange tax

Most developer tools bill in USD. If your revenue is in AUD, every invoice carries currency risk and a conversion spread, typically two to three percent on a standard business card, sometimes more.

Two or three percent on a small hosting bill is nothing. On a meaningful infrastructure spend it's a real number, and it moves without warning when the exchange rate does. Budgeting in AUD against USD-denominated costs means your infrastructure line item varies by ten or fifteen percent year to year for reasons entirely unrelated to your usage.

There are also accounting consequences. An AUD invoice from an Australian entity comes with GST handled the way your accountant expects. A USD invoice from an overseas supplier is a different treatment, and reverse-charge GST on imported services is a conversation worth having once rather than discovering at tax time.

Where local vendors exist and are competitive, SMS is the clearest example, billing in AUD from an Australian entity is worth more than the headline price difference suggests.

Where they don't, the mitigations are ordinary: a card with low foreign transaction fees, or a multi-currency account holding USD if your spend justifies it.

Our category pages record billing currency and entity per vendor so you can filter on it rather than discovering it at checkout.

Frequently asked questions

Do I legally need to host Australian user data in Australia?

Usually not. The Privacy Act's cross-border rules are largely about remaining accountable for how overseas recipients handle the data, which contracts can satisfy. Government, health and some financial services work carries harder requirements, often contractual rather than statutory.

Is Singapore close enough?

For many applications, yes, 90 to 100 milliseconds is tolerable for non-interactive work. For anything conversational, or anything making several sequential round trips per page, a Sydney region is noticeably better and increasingly available.

Which providers have Sydney regions?

AWS, Google Cloud, Azure and Oracle all do, as do Fly, Vultr, Linode and several object storage providers. Cloudflare has multiple Australian edge locations. Our category pages record Australian region availability per vendor.

Should I use an Australian vendor?

Where one is competitive, yes, AUD billing, local support hours and an Australian entity are worth real money. In SMS the local providers are cheaper as well. In storage and compute the global players usually win on price and features even for Australian workloads.