Wednesday, July 10, 2019

The future of the NBN; one year to completion

As we start 2019/20, the 2020 end of the NBN rollout is now very close. The weekly NBN progress report shows 8.1M brownfields premises are Ready to Connect (of total 9.8M), and overall 5.5M premises are connected. So now, our thinking turns to what next after the rollout is complete.
Jeremy Segrott, Flickr (CC-BY)

There will still be:
- getting all the ADSL network shut down,
- getting as many premises as possible on to the NBN, not least in trying to find the right combinations of price and service to suit the diversity of Australian households
- trying to earn a profit from the completed network, and lastly
- thinking about what next.

Three data points, I came across this week, give pause for the what next question.

1. NZ quarterly report of their NBN; the UFB;
While NBN reaches 5.5M activated premises (over 50%), Chorus in its 2018 Annual Report advises that their activations have risen from 35% to 45% (p.1). Across the whole of NZ, UFB Quarterly Report shows their activations include 8% on 1Gbps (64,000 connections), and 6.5% on 200Mbps (49,000 connections), with the remaining 85% (645,000 connections) on 100Mbps or less.

What next for NBN includes when to activate 1Gbps services, and at what price.
The website, shows NZ services can get 1Gbps service for about $100NZD per month, with unlimited data, 12mth contract, and $150 for a modem/router. If NZ can only get 10% of connections onto gigabit at $100 per month, then don't expect Australia to reach 10% gigabit unless it is priced affordably and competitively. That is, at $600 per month for gigabit, expect only low single digit percentage household uptake. There may be more interest from businesses at this price.

NBN says 50% of the network will be gigabit compatible, including HFC, FTTC and FTTP. With Gigabit worldwide reaching only in 2017, Speedtest reports only around 400,000 users. Australia will soon have 4M capable NBN premises. But whether they will use gigabit will depend on need and price - hence the value of gigabit.

Chorus (the wholesaler) in their 2019 half-year presentation, say their gigabit charges are $60 per month for Residential and $75 per month for Business connections

2. Speedtest compares Australia, Saudi Arabia and Ireland
While the analysis of the three countries was not that interesting, a graph of Australia's broadband was very interesting. Speedtest provided a comparison of AU broadband speeds in 2018 and 2019 speeds in a graph at 5Mbps intervals. Their testing shows no measureable  usage over 100Mbps. Each 1% equates to about 100,000 households. There are obvious NBN peaks just under 50Mbps and 100Mbps,  and obvious gains in those peaks between 2018 and 2019 (see the Data).

The graph shows under 25Mbps falling from 70% in 2018 to 50% in 2019. In the same time average speed increases from 30??Mbps to 40??Mbps. If all the under 25Mbps were to be removed, then the average for Australian would be around 50Mbps. This does not take account of underperforming NBN services such as congested fixed wireless.

This data and graph shows Australia is quickly improving. But compared to the top 20 nations who are averaging close to 100Mbps, we are well behind. Not until gigabit takes hold will the Australian average leap forward. The NZ UFB (per the 2019 Q1 report) averages around 170Mbps with the 8% gigabit having a big impact on lifting the average.

3. OECD compares member countries by broadband speed, including per cent of connections over 100Mbps.
OECD data (xls)  on broadband speeds for 2018 has Australia on 0% of connections over 100Mbps. This puts Australia, next to Greece at the bottom of the OECD. In total about 70M connections over 100Mbps are noted across 37 countries. The US has about 30M of these connections (about 10 per 100 people). The next biggest countries being Germany (4.9M), Italy (4.6M), UK (4.3M) and Spain (4.0M), at either 15% (DE, UK) or 25% (IT, ES) of broadband connections. By percentage the most over 100Mbps connections are Switzerland (84%), Sweden( 67%), Portugal (63%) and Belgium (52%), but in total these countries only have about 11M total connections over 100Mbps, about 2-3M each country.

There is also comparative data ( | xls) from 2016, so it is possible to see how fast people are adding over 100Mbps connections. And now I also retrieved 2014 data from the OECD Digital Economy Outlook 2015 (Broadband speed data (xls); Figure 2.26, p.109).

Monday, February 18, 2019

The value of non-profit: Dryad and California Digital Library

Workshop: Accelerating Data Publication: new models for research institutions. Melbourne 8 Feb 19.

I attended a workshop the other day, where John Chodacki, Director, UC Curation Center, California Digital Library, pitched an idea for a Data Repository, run by the non-profit organisation, Dryad. Dryad, accepts researcher's data, for (as John mentions) around an average $120USD processing fee, which covers the curation cost to check all submissions (data files and data description), and to hold the data long-term to support a publication related to the data. Here are the slides and Etherpad. Per Dryad's website, the charge per publication is calculated as a $120USD processing fee plus a storage fee if a researcher submits over 20Gb data ($50 per 10Gb). Most datasets come under the data cap, John says. This fee is waived, upon request, for researchers from lower  - middle income countries (per World Bank definitions).

John had another idea, for how to use Dryad, for his University of California (UC) researchers. UC had been considering building their own repository from scratch, but thought their funds might be better spent building on top of an existing platform - where researchers were already depositing their data. Apparently 90,000 researchers (Slide 55) already have deposited data at Dryad, and Dryad provide data deposit for over 900 academic journals (Slide 55).

John's idea was to use Dryad as an Institutional service, where the University could pay an annual fee, then their researchers would no longer need to pay a per article data deposit fee. Talks with Dryad began, and they agreed to seek support of other institutions to promote the new service. Fees were agreed on a sliding scale, where a big Institution would pay more (about $13,000 USD per year) and a smaller Institution would pay less (about $3,000 USD per year).

At the workshop I attended, there were representatives from a number of Australian Universities and research organisations; RMIT, Deakin, Victoria University, CSIRO, University of Sydney, University of New South Wales, Queensland University of Technology, Aarnet, Monash University, plus myself and my colleagues from ARDC, the Australian Research Data Commons. In a final quiz item, the attendees said they were interested in the new service, but that they were not the ones to sign off on that type of purchase. It raised the question of what was the next steps. How could Australian institutions contribute to interest in this new idea for a repository? This blog post was one of those next steps.

There were a number of obvious benefits:
- paying a non-profit community run organisation, rather than a commercial provider, so it should be a little cheaper
- community can drive what the system looks like, and the functions can be responsive to community needs
- using a system that already has a lot of researcher users (over 60,000), and journal integration.
- using a trusted system (certified with CoreTrustSeal recognised certification).

But there were some obvious challenges:
- if you were to use this system, how would we transition from our current system to this new system?
- would there be local support in Australia to help users, and help Institutions make a business case, and help researchers use the Dryad system?

There were a number of questions raised:
- where would the data be stored? In the US. But a copy could be kept locally.
- what about sensitive data, which has to be stored in Australia (or an equivalently privacy-protected jurisdiction ie not the US which has weaker privacy rules)? Dryad doesn't accept sensitive data, so where the data is stored should not be an issue.
- what size data can be sent to Dryad? (As a file up to 10Gb; and from a URL up to 100Gb)
- can we use specific data description types (vocabularies)? Should be fine, John said.
- can we get descriptions of data belonging to our Institutions, back into institutional systems? Contact Dryad for discussion.
- have CDL/Dryad talked to the Australian infrastructure provider, QCIF, who manages the RedBox repository? or Figshare? about how to create a win-win amongst the various repository providers? No, but there is enough room in the ecosystem to allow different repositories to service different users/
- were there any plans to have people in Australian to promote the services and educate institutions and users about the benefits of the system, like Figshare's sales team? Not at this stage.

Later that same day, I was at another event, and spoke to Natalie Meyers, University of Notre Dame, who is an ambassador for the Center for Open Science and interested in having more integrations with the Open Science Framework (OSF). She wondered if by engaging a larger community to join a Dryad Institutional service, there might be incentive to enhance the Dryad API to make it more consumable and its endpoints more interoperable with other open science platforms like OSF? John, in response, says the new software development work  at CDL/Dryad will provide exactly this type of opportunity by leveraging CDl's DaSH API (see API docs and github repo). 

So, it is obvious that the community needs to keep talking about what to do next, what questions are still left unanswered, and how to sell the idea to the money people at your institution. That will require at least in the beginning, glossy flyers, and ultimately a business case to assess the costs and benefits. See also the Dryad blogpost (May '18) announcing the partnership with CDL (follow up Oct '18 post).

If you have any questions, please post them in the comments section below and I will send them along to John (john.chodacki - at- to answer.

NB: Where I work at ARDC, we have a policy of vendor neutrality, so this post should not be read as an assessment of the value of the CDL/Dryad service, but does seek to point out the many issues to be wrestled with in making a value assessment.
Thanks to Natalie and John for their feedback on this post.