ICT-KM of the CGIAR

Collaborate, Create, Communicate

Is our bandwidth sufficient to do networked science?

This article was first published in the International Livestock Research Institute’s (ILRI, www.ilri.org ) internal intranet site. ILRI hosts the Bioscience facility for East and Central Africa (BecA) and is ramping up its work in the area of biotechnology. However, East Africa will be the last region in the world to be connected to the internet by optic fibre cable. Slow, expensive, often contested satellite bandwidth is a big constraint to carrying out this work. These are some thoughts from our partner Erik Bongcam Rudloff and from Etienne de Villiers and Ian Moore of ILRI.

———————————————

Is our bandwidth sufficient to do networked science? and if it’s not, what are the implications and potential solutions? Ian, Etienne and Erik puzzle it out!

Ian, Etienne and Erik
Ian, Etienne and Erik

Today’s scientific work is becoming more networked!

Erik Bongcam-Rudloff, associate professor of bioinformatics at the Swedish University of Agricultural Sciences (SLU), talks about networked science and the need for high bandwidth.

Erik Bongcam-Rudloff
Erik Bongcam-Rudloff

Erik, who has been a regular visitor to ILRI Nairobi since 2006, says: ‘Scientific work is becoming more and more networked and technology development is going VERY fast.

 

‘We are in a golden era of development. We need to look at bandwidth and the channels (ports) we have open for communications and what scientists need to ‘do’ science today and tomorrow.

 

‘Email is just a primitive communication tool from the past!

 

Erik argues: ‘Good internet connectivity is more imperative than ever before.

 ‘You’ll soon have a new 454 machine at the BecA-ILRI hub. [For non-scientists a ‘454’ is a ‘second generation’ high throughput sequencing machine – apparently the ‘new paradigm in sequencing’!].

About Erik
About Erik

‘These machines will change the way people design experiments and allow us to ask research questions that weren’t possible before. These machines can produce lots of data… but without knowledge the data is worthless! It’s all about how you connect data to knowledge and no single institute or university or lab is capable of coping with this alone. New science demands that many people – many thousands of people – connect and share in real time!

‘So there’s a huge need for a very fast internet connection to distribute, communicate and do analysis of scientific data. All future research will be done collectively. Living in silos, unconnected to the rest of the world, is out!

Fibre optic connection coming soon! 

Ian Moore
Ian Moore

Ian Moore (ILRI-ICRAF ICT Manager) says: ‘Reliability, bandwidth capacity and speed are the three most important factors that we take into account when implementing an internet connection. A research institute the size of ILRI should have an absolute minimum capacity of 10Mbps of uncontested bandwidth on each campus, preferably through a fibre optic connection to the internet.

‘This capacity is still small when compared to universities like SLU where Erik works or even compared to our sister centre IRRI in the Philippines.

‘In Ethiopia the optic fibre internet connection is fast but very unreliable so we’re going to install a satellite connection to improve the reliability, but unfortunately that won’t help the speed!

In Kenya and many other parts of eastern and southern Africa we’re still waiting for fibre optic internet connections, so until then we have to access the internet through slower, expensive satellites. For an asymmetrical satellite connection of 4Mbps incoming and 1Mbps outgoing in Nairobi ILRI spends an extortionate USD 180,000 per year. In Europe you can have shared 10Mbps in your house for USD20 per month!

The growth of the mobile phone industry and the loss of several satellites means that available capacity in the region is limited… costs are high and our upgrade options are limited. The limited bandwidth means that the link becomes congested at peak times and this is the main reason for slow speeds.

‘But the good news is that fibre optic internet connections from TEAMS, a Kenyan Government project, and SEACOM will land in Mombasa by March and be commissioned in June [That’s THIS year: 2009!]. The cost of bandwidth is estimated to drop to around USD800 per Mbps per month. If ILRI continues its policy of increasing bandwidth rather than making savings when prices fall then we’ll be able to implement the minimum 10Mbps internet connection that we need. Fibre optic connections are 5 times faster than a good satellite connection, so speeds will improve too.

So what can we do in the meantime?

Ian says: ‘First we needed to make sure that we were using our bandwidth productively. We’ve been monitoring to ensure capacity is not being taken up by viruses, other unexpected traffic, that staff are not using the internet for personal entertainment or gain and that we’re caching regularly accessed content. But at the same time we don’t want to frustrate scientists by blocking access to sites they need to do their work!

‘We’re now confident that the large majority of bandwidth is being used solely for work-related purposes. We’ve also ordered a bandwidth manager device which will be installed within the next month. This will give us more flexibility to assign priority use of the bandwidth to specific groups of users or to certain types of internet traffic. The young scientists who struggled at the bioinformatics webinar last week will, in future, receive the bandwidth they require, but this will be at the expense of others. These measures offer only limited respite and soon we’ll have to upgrade our bandwidth capacity.

‘I totally sympathise with the scientists, especially the bioinformatics team who need to regularly update their huge datasets. If we were located in a region with fast internet 2 connectivity for research and education establishments, like GEANT in Europe or APAN in Asia, then Etienne (de Villiers) and his team would be able to download a dataset within a matter of hours. At the moment, it’s quicker for them to receive datasets on removable media via DHL, rather than attempting to download them through our internet connection.

To cater for the new generation of scientists, make use of the new communication and collaboration tools, not to mention the resources required to carry out research on the internet, we DO need more bandwidth. So ILRI scientists and management have to weigh up if the savings and improved productivity that can be gained from a non-contested internet connection is worth the increased investment in bandwidth and whether this investment should be made immediately or whether ILRI can afford to wait for the faster fibre optic cables to be commissioned.

‘The existing internet connection is funded almost entirely from unrestricted core funds recovered through the ICT service charge. But this is not sustainable and in the future more funds need to be built into restricted grants especially by those who need the additional capacity. So we need to do an assessment with scientists and figure out the best way to go.

Is it possible to ‘stay in the game’ if we don’t have high bandwidth?

Erik believes that productivity improvements and other costs savings can be made by upgrading now: ‘Huge amounts of money can be saved by increasing bandwidth. I’m chairman and board member of two international bioinformatics groups. We meet once a year face to face, but hold monthly meetings over the internet using a webcam, microphone, freely available video conferencing tools and a good internet connection. This saves us at least Euros 100,000 a year in travel costs AND reduces our carbon footprint! We talk to people in China, Brazil, South Africa and all over the world. We can have rapid questions and answers and this saves us weeks, if not months, of time! We also give teaching courses through this system.

I believe the future of science is in building gigantic wiki-like systems where whole communities collectively write datasets and these datasets will, of course, be open source!

‘We already have two examples – Wikigenes and BioGPS. It’s amazing how much data is there already. Yes, it’s primitive at the moment but so was Wikipedia when it started and just look at it now! There are few scientific articles today that don’t cite [our friend] Wikipedia as a source! [Tip for scientists: Check out Wikipedia’s page on your research topic and make sure your research is cited! And if there isn’t a page – just create one!]

Erik concludes: ‘So is it possible to do research with the latest technologies if you don’t have high bandwidth? No, it is not possible any more! No research can work by itself. Today’s science is network-based using a plethora of internet-based tools.

Scientists need to be prepared for collective working. And a lack of bandwidth will hinder real progress. If you don’t practice and use the tools, then you will be left behind! This new way of working is here now and it’s already ongoing.’

So do you agree with Erik’s views on where science is going? What are your experiences? Can you provide examples of this new networked science and how it’s working in your area? Please post your comments below.

—————————————————–

Article by: Margaret Macdonald-Levy with thanks to Erik Bongcam-Rudloff, Ian Moore and Etienne de Villiers.