Last year, the Square Kilometre Array (SKA) broke ground in Western Australia. A signature event for radio astronomy. Over 20 years in the planning, the SKA is a worldwide collaborative science project, with investments in South Africa as well as Australia, to make a giant radio telescope (there’s a cute trick in astronomy where two simultaneous observations widely separated on earth is like having one telescope with the ‘size’ of the distance between them).
So… what’s interesting about this from an Internet operations perspective? Well, two things.
Firstly, it’s going to make a lot of data. A lot. Initial design estimates suggest an exabyte of data per day, which means a massive investment in academic and research networking to ‘sink’ that amount of traffic every day, without end. And that, in turn, will have implications for connectivity from Australia (in this case from West Australia) to the broader world.
An exabyte is 2 to the 60th — one billion, billion. At this point, anyone who works with IPv6 is smirking because the maximally routable prefix in IPv6, a /64, is sixteen times larger, regardless…
Will this huge scale of data be an issue for the modern 2020s Internet? Luckily, the answer is no, for multiple reasons.
Firstly, the academic and experimental networks space has always been mindful of its impact on the bottom line because the budget implications of data networking have never been far from the surface in a year-by-year grant review world. So the cost of working with these petabytes and exabytes of data has to be understood by SKA’s data scientists. SKA architects will have an acute understanding of peak and sustained data needs and if the network looked like it was going to saturate, re-scaling costs would be built into the model.
Secondly, academic and experimental network networks use commodity protocols (Internet Protocol and related standards) but aren’t necessarily dependent on general Internet access. Unlike regular user web access, dedicated links will be provisioned where necessary: Domestically, to get data from the SKA array to onshore data processing facilities, and internationally, to pass data into the wider research community. Just as global communications, in general, has moved from long-distance transit to an increased dependency on local caches, the cache managers and cloud providers have been moving their Internet traffic to private links: Google now routinely deploys it’s own trans-oceanic fibers routinely deploys its own trans-oceanic fibres.
Lastly, it’s very likely that astronomy data can handle a little bit of data loss in the short term. The reason is quite simple: The sky will look mostly the same tomorrow night. There are certainly transient events that happen in an instance, the most famous being the ‘Wow! signal‘ in 1977 that lasted just 72 seconds. But this isn’t the main target of SKA research, it’s going to be used to look into deep-time, far-away radio signals, which are likely to display long-term persisting trends.
So if network protocols that can handle a little bit of loss are used, like a UDP file synchronization system, it shouldn’t present a big problem. This means that the network can be designed to handle surges in data reasonably ‘gracefully’ for other uses.
This isn’t the first time the Australian Internet has been massively influenced by astronomy and space science — the birth of the Australian high-speed connection to the Internet lies in a satellite link created between the Australian research community and the NASA Science Internet to facilitate communications between various US research programs and Australian research institutions. The use of this connection included aspects of radio astronomy but was intended to be a general-purpose connection used by researchers for all types of research.
Prior to this, Melbourne University had been running a dial-up link to the US, which carried messages, mainly in the form of email. There were, of course, other dial-up links in Australia at that time (in the late 1980s), but few of them had any global context. In contrast, the Melbourne University-operated dial-up link was being used to connect the CSIRO and all of the Australian university computer science community and an emerging community of Unix users for email and network news.
The relay-based messaging network worked moderately well, but it lacked the immediacy of a direct network connection, and by the end of the 1980s, it looked like the stars were aligning to bring the technology and the money together to build a national academic and research network that featured a permanent connection to the national research and education network (NREN) in the US.
It seems to be a constant feature of networking that there is always more data than there is network capacity to carry it. The initial circuitry of the Australian Academic and Research Network (AARNet) was based on digital voice circuits, with a 48Kbps capacity. The satellite circuit was a 56Kbps circuit. This was quickly saturated and the ensuing years were an endless cycle of chasing more money to pay for larger circuits and chasing higher-speed transmission services to carry data at these higher speeds.
The network quickly moved from kilobits per second to megabits per second, then to tens and hundreds of megabits per second. The next generation cycled through gigabit-per-second fibre technology and no matter what was installed, there was more than enough data moving around to fill up the network.
Even now in the era of terabit per second networks, which, from the perspective of the late 1980s is an impossibly large capacity, ever-larger applications are dreamt up that can use this network and Big Science has made some other notable contributions to this work of combining big data with big networks, including the Large Hadron Collider, orbiting telescope programs, and SKA.
In truth, the roots of network linkages to astronomy are even more profound, dating back to the CSIRO Radiophysics division’s birth in 1940 and the invention of radio astronomy by, amongst others, ‘Taffy’ Bowen. Admittedly, this was long before the Internet was a dream, let alone a concrete protocol. Nonetheless, the Australian computing and networking community has astronomy at its core.
The views expressed by the authors of this blog are their own and do not necessarily reflect the views of APNIC. Please note a Code of Conduct applies to this blog.