Saul Albert

Research on aesthetics, technology and social interaction

Archive for ‘July, 2016’

Syncthing Logo

One of the major logistical problems facing researchers who use large audiovisual data files featuring recordings of human subjects is how to share it with colleagues simply, securely and inexpensively.

  • By simple, I mean that the a/v file-sharing solution should not need specialized equipment or an expert systems administrator to set it up for you. Most universities and research institutions have in-house research file servers. These may be security audited, up-to-date and well-organized – they may seem unlikely to suddenly lose or delete all your data or randomly restrict access to your colleagues in other institutions. However, in my experience, it’s best to manage your own data and backups!
  • By secure I mean that it should not rely on cloud-hosted storage (which may store and transfer data anywhere in the world), or allow data to be transferred unencrypted between remote systems. This is often a requirement of UK human subjects/ethical approval – so it’s particularly relevant in the case of UK educational institutions, but this is probably also true elsewhere.
  • By inexpensive I mean that it should not cost the end-user an ongoing fee, or meter use per gigabyte stored. Commercially available cloud file-storage services like Dropbox, OneDrive, Google Drive or iCloud are relatively cheap up to ~ 1TB storage, they can get quite expensive beyond that, and if you have multiple projects with multiple researchers in different groups sharing data – the total cost for all researchers can become prohibitive.

There are trade-offs between these three requirements – this blog post outlines how to use Syncthing – which I think is an optimal solution – to solve this issue.

What is Syncthing?

Syncthing is a file synchronization tool – much like Dropbox, but it is peer-to-peer, which means it works like Bittorrent or other file sharing tools that do not require a central server-based system to share files.

A client-server model (left) and a peer-to-peer system (right)

A client-server model (left) and a peer-to-peer system (right).

Why use Syncthing?

Firstly, Syncthing is an open source, peer-to-peer file sharing system, which means it is relatively cheap, simple to set up and secure. It is especially good for sharing very large files and collections of files without having to pay or to trust an intermediary to maintain a centralized file server. You run Syncthing on your computer, the person you want to share files with runs it on theirs, and you can set up folders that will synchronize files automatically when both your computers are turned on and connected to the internet. The data is encrypted in transit, and never sits on someone else’s server. Syncthing only requires each user to have a normal computer or laptop, rather than running on a server with each person using a ‘client’. This is more secure, and probably less complex to set up and maintain.

Secondly, Syncthing is open source, under active development and compatible with all major computing platforms because inevitably people will use Mac OSX and Windows, and sometimes *nix. Syncthing is one of many open source tools relevant for educational contexts, which are not only cheaper for individual researchers and teams, but also tend to stick around for longer.

Finally, Syncthing allows each user to choose how they want to organisze their folder structure. Whereas Dropbox uses a standard ‘dropbox’ folder, and (by default, at least) forces everyone use the same folder structure and folder names for their files, Syncthing allows you to put your data wherever you want on your hard drive, but still choose it as a folder to synchronize with your colleagues. So I may choose to put my video file in a folder on my Linux machine here:

/home/saul/data/project1/video1/video1.m4v

while you might have your videos on your windows machine at

C:\Users\Yourname\projects\project1\videos\video1.m4v

A third colleague may store video on their desktop (tut tut) on their Mac at:

/Users/user/Desktop/video files/video1.m4v

We can all then use Syncthing and choose to synchronize my ‘/video1/video1’ folder with your ‘\project1\videos’ folder and our colleague can synchronize with their desktop ‘video files’ folder – so despite us all having different data naming schema, we can collaborate effectively, share files, and keep our own file systems organized in a flexible and personalized way.

For many years I’ve used Dropbox but I’m always running out of space, then having to weed out awkwardly placed or duplicated files from collaborative projects which may or may not still be used or needed by collaborators. So this feature of Syncthing is a huge selling point for me.

What is Syncthing not so good for?

Syncthing is not particularly good for synchronizing millions of small, regularly updated files. Syncthing monitors which files need syncing by scanning its folders every few seconds to find out what has been updated – this can be a bit processor intensive if you have hundreds of thousands of files to scan through. So, it’s best to use it for projects with a few thousand large (especially video/heavy data) files rather than projects with tens of thousands or millions of files. For the same reason, if you need the files to remain synchronized in (close to) real-time, Syncthing’s folder scanning process will probably take too long.

Syncthing is not great for always-online files. Because it is a peer-to-peer system, Syncthing requires the computers you are keeping in sync to be online at the same time for syncrhonization to take place. Dropbox, by contrast, uses an always-online server, so it doesn’t matter if your multiple computers running Dropbox are online at the same time or not – the server will make sure they all have the most recent version of your Dropbox files. So, if you need something always-online, better to use a server-client system like Dropbox, OneDrive, iCloud etc.

Syncthing is not particularly useful on smartphones/tablets. Finally, although Syncthing does have an Android client and an iOS client, these are not officially supported by the same developers who work on syncthing, nor are they going to work if your computer is off-line when you need to grab a file via your mobile syncthing client.

Installing Syncthing

Syncthing has excellent, up-to-date documentation that will guide you through the installation process. However, there are also several helpful videos that will help you install Syncthing on different operating systems. I couldn’t find a video about how to install Syncthing on Mac OSX, so I made one myself.

Installing Syncthing on Mac OSX (with Homebrew):

Installing Syncthing on Windows

Installing Syncthing on Linux

Setting up and using Syncthing to share data

If you want to learn how to set up Syncthing for special purposes, and you want to explore all the options, I recommend reading the documentation, and if you want an in-depth video guide, I recommend the Nerd on the Street guide to setting up Syncthing. However, for a quick-start video, once you’ve installed it, I’ve created a short video that shows you how to use Syncthing to keep a folder in sync between two computers with the simplest set of default options.

I’ve used Mac OSX (el capitain) in this video, because I think it’s actually harder for most OSX users to understand how to fill in the Folder Path options (the location of the folder you want to sync) when setting up Syncthing for the first time. It’s relatively straight-forward to find Folder Paths in Windows and if you’re using Linux, I expect you’ll already know how to do that.

Troubleshooting

One major issue I’ve seen people having when running Syncthing for the first time is getting Syncthing to run at startup automatically. Although it’s beyond the scope of this article to deal with application startup issues – I’m happy to offer advice with this if you leave questions in the comments after reading the documentation.

Another major issue I had to deal with was cross-platform file-naming issues (especially for Mac OSX users). Basically, different file systems (FAT32, NTFS, exFAT, UFS/+, ext2/3 etc. etc.) allow different kinds of file names. If you are synchronizing folders between different file systems (on external drives, system hard drives, different operating systems etc.) it makes sense to use very conservative conventions.

I recommend the following:

Alternatives to Syncthing

There are many alternatives – some of them look very interesting, but none of them fit all the requirements outlined above as well as Syncthing. I’m open to adding to this list – so if you have a solution that isn’t shown here, please do let me know.

Finally – a very good solution, with minimum fuss or technology is to just copy your files onto small external hard drives and snail-mail them to each other in well-padded mailing boxes. That’s often a simpler (if slower) solution than setting up one of these systems!

Related:

Leave a comment

I just ran a 2 hour workshop for the Media and Arts Technology Doctoral Training Centre using a variant of the Spectroscope Facilitation method during which 10 people showed and discussed their researcher / artist / technologist portfolio websites and gave each other structured feedback. After the show & tell session, each of the participants went home with 27 prioritised ‘todo’ advisory notes written by other participants to improve their own websites. However, one of the most useful outcomes – that I’m going to share here – was a set of 10 best-practice tips derived from each participant writing down what they really liked about each other’s sites.

Spectrascope facilitation cards

These are neither comprehensive, nor all compatible with one another – and I certainly could take much of this advice to improve my own website, but anyone thinking of setting up and academic / personal / portfolio site might find these useful. The most important question for anyone to address with their website was to ask who and what is the website for. Is it a CV to convince potential employers in academia, or to appeal to ad agencies, or to attract artistic commissions? Once these central questions were answered, the following tips could be used to improve each site.

  1. Prioritise clarity and coherence.
    • Make a clear statement of who you are and what you do in plain language – right up front.
    • What is the website for? If it’s basically a CV, use the word ‘CV’. If it’s a portfolio, use the word ‘portfolio’. Shape the expectations of the visitor so they know what they’re getting.
    • Add this information to the title/metadata the homepage, so the title bar (and Google’s web crawlers) read e.g. “Name, Job description, Purpose of site”.
    • If you have a lot of projects to show, put a selection of only the best/most targeted to the purpose of the site on the front page.
  2. Foreground your main activity/role.

    • If you have multiple quite different audiences/purposes in mind for your site, consider making multiple sites.
    • However, it can also be useful to show multiple facets of what you do – if one is a minor activity and the other a major activity, combine and show a richer combined picture.
  3. Keep it current.
    • Make sure you can update it easily, reduce friction by choosing technology that you enjoy using.
    • Show what’s upcoming – tell people what you’re doing next.
    • Avoid showing out of date information by e.g. not using temporally-relative wording ‘this year’ etc.
    • Keep your CV rigorously up to date and centralised (i.e. don’t have Linkedin profiles or freelance CVs sitting out of date around the web). A good hack for this is to use a Google Doc that you update regularly – and link to a PDF-downloadable version of it from your website.
  4. Show people where to go.
    • Show as many of your menu items as you can all at once – don’t hide menus behind fancy dropdowns or multiple click-throughs.
    • Optimise for ‘fewer clicks’. Try to make everything on the site only one or maximum two clicks away. The most important things must be one click away and immediately visible on the homepage.
  5. Get a domain name.
    • Buy your own domain name. It costs almost nothing and it mostly looks more professional.
    • It’s also more future-proof, i.e. if you buy www.myname.com and point it to a wordpress site, or a squarespace site, or a cargocollective site, but then change where you host the site, you can take the domain with you. If you relied on wordpress.com/myname – you’re stuck with wordpress.
    • If you can, use short urls e.g. http://myname.com/about or http://myname.com/project/myproject – this is relatively portable (as above) so you can change which software you use to produce your website without being tied to software-specific page names and urls.
    • It’s good to have your own domain for email (especially if you’re a freelancer – less an issue in academia where people use institutional emails). However, be careful not to use vanity emails like info@myname.com to sign up for core services. Here’s why.
  6. Use a nice photo of yourself doing something.
    • It’s good to show something about yourself – but try to show yourself doing something relevant to the style and purpose of your site.
    • You might want to use a widely identifiable gravatar so that your website is visually identifiable with your social media profiles/comments from around the web.
  7. Use relevant copyright notices.
    • As long as you’ve created your website, whatever you publish is legally yours whether you add a colophon with a ‘©’ symbol or not. The symbol is just there to advise people on how to use your content.
    • If you are an artist with lots of great visual work – which unscrupulous people love to use without permission – say how you want people to use it.
    • If you want people to spread your work and give you credit, use a license such as the CC-BY – or use the CC0 option if you want to put your work in the Public Domain.
  8. Offer alternatives to audiovisual content.
    • A few images work as well as a video, sometimes better.
    • Consider using a simple explanatory animated GIF so people don’t have to download a whole video. 1
    • Always consider people with low bandwidth, small screens, out of date browsers. Responsive designs are easy these days with lots of great templates available for most website/blogging engines or just HTML5 templates.
  9. Enhance your ‘findability’.
    • Add a list of hardware / software / techniques / approaches used for each project. This shows what you can do, and it helps people using Google to find you and what you know about.
    • In general think of your site as a dragnet for people to find you. It’s much better to be found than to go knocking on people’s doors – so think about who do you want to find you.
    • Use analytics – you should know how people see your site, where from, and which pages they visit – it helps you make decisions about how to change and update it.
    • Link to your social media/other platforms from your site and vice versa (linkedin, twitter, github, academia.edu, researchgate, your institutional sites etc.)
  10. Show people who you are.
    • It’s amazingly important to care about your web presence these days – make it reflect who you are, what you care about and believe, and make it unique. Of course this also means being critical, self-aware and careful not to project those bits of yourself that might undermine the purpose of your site!
    • Sometimes a web 1.0 site at an address like  http://institutionalname.edu/~yourname is a good way to show who you are – if you’re an academic in engineering or computer science. Go with that, engineering academics who are looking to hire you will recognise you as one of their own.
    • A really beautiful, unique and intriguing image of your work – or a great video or poem can be a wonderful hook into an art or design-focused site. Intrigue people, then reward them with more eye candy and carefully thought through information.
    • Show your network – link to others, make sure to credit all collaborators and link to them, they’ll appreciate it!

These are somewhat general tips. There was also a lot more technical advice in our email thread about this workshop, as well as some advice on which website services might be useful. Some of those links are included below – but any further explanation as to what to do with them is well beyond the scope of this post!

Beginners:

  • WordPress – either as a hosted service – or self-hosted on your own server. If self-hosted, beware! It can be tricky to maintain and has had lots of security problems in the past.
  • Squarespace – looks easy, but it’s pretty expensive for a simple portfolio/website.

Intermediate/advanced: (thanks to Victor Loux).

Hosting:

Thanks to Toby Harris, Victor Loux, Daniel GabanaJacob HarrisonJulie Freeman, Laurel Pardue, Raphael Kim, and Betül Aksu for presenting their works in progress and giving great feedback.


Update 7/18/2016:

Travis Noakes suggests (see the comments section below), using a unifying visual metaphor that brings together your website, your visual presentations and even the binding of your thesis. Travis’ excellent research blog uses the “+” sign to do this and is well worth checking out as a great example of a researcher’s site that integrates his description of his roles and foci with the site’s navigation and visual communication. He also suggests using Google’s Blogger platform for enhanced google juice and ease of use.

Related:

Notes:

  1. A good tip on this from Victor Loux: If you’re considering using a GIF to show a specific interaction in a project, also consider that GIFs can be several megabytes big if they’re wide, as well as being lower quality (limited to 256 colours and less fluid). The trick I’ve used for my website (the ‘PeDeTe’ project) is to actually use a <video> element that acts like a gif (autoplay, looped, and no sound); for the same video length and same resolution, it reduced a 4.8 Mb (!) GIF to a 395 kb mp4 file. Most modern browsers support it and you can certainly find/make a polyfill for older ones. The only downside is that iOS will refuse to autoplay it, unlike GIFs, so that’s just not a viable option if mobile is really important.
2 Comments