Pycon Italia 2017: Planet Friendly Python

My talk from Pycon Italia Europe about Planet friendly Web development with Python.

Transcript

1. Planet friendly
   Web Development
   with Python
   Chris Adams
   Pycon Italia 2017
   

   View Slide

2. Chris Adams:
   @mrchrisadams
   This guy
   Environmentally focussed
   web generalist
   

   View Slide

3. A.M.E.E. (Avoid Mass
   Extinction Engine) - CO2
   calculation APIs
   Loco2 - Low CO2 travel
   in Europe by train
   TeaLight - Low CO2
   cups of tea, with
   demand response
   My Background:
   

   View Slide

4. Pycon Italia 2017 - Planet Friendly Web Development
   @mrchrisadams
   

   View Slide

5. Pycon Italia 2017 - Planet Friendly Web Development
   @mrchrisadams
   @mrchrisadams
   What can a technologist do about climate change: worrydream.com/climatechange
   

   View Slide

6. Pycon Italia 2017 - Planet Friendly Web Development
   @mrchrisadams
   http://www.iea.org/media/news/2017/GlobalCarbonEmissions2.png
   http://www.iea.org/newsroom/news/2017/march/iea-finds-co2-emissions-flat-for-third-straight-year-even-as-global-
   economy-grew.html
   

   View Slide

7. Pycon Italia 2017 - Planet Friendly Web Development
   @mrchrisadams
   http://folk.uio.no/roberan/t/global_mitigation_curves.shtml
   

   View Slide

8. Pycon Italia 2017 - Planet Friendly Web Development
   @mrchrisadams
   @mrchrisadams
   World GHG Emissions in 2010 - Analysis by ECOFYS
   Source* Greenhouse gas
   Coal
   25%
   Natural Gas
   19%
   Oil
   21%
   Waste
   0.4%
   Sector
   Direct
   emissions
   34.6%
   * Greenhouse gases can arise from two sources
   N20
   7%
   CH4
   15%
   CO2
   76%
   HFCs & PFCs
   2%
   Transport
   15%
   Residential Buildings
   11%
   Commercial/Public Buildings & Services
   7%
   Land Use
   Change
   15%
   Industry
   29%
   7%
   6.0%
   4.8%
   4.3%
   1.4%
   1.1%
   1.0%
   10.5%
   1.5%
   2.6%
   10.5%
   5.5%
   1.4%
   1.8%
   3.1%
   4.4%
   10.3%
   Other industries
   Paper, pulp and printing
   Food and tobacco
   Non-ferrous metals
   Chemical and petrochemical
   Iron and steel
   Non-metallic minerals
   Aviation
   Others
   Road
   Livestock and manure
   Agricultural Energy Use
   Coal Mining
   Oil and gas extraction, refining and processing
   Deforestation/Afforestation and Land Use CO2
   Agricultural soils
   1.3%
   1.6%
   Landfills
   Waste water & others
   Waste 3%
   Agriculture
   Energy Supply
   13% 8.3%
   Energy industry own use & losses
   

   View Slide

9. Pycon Italia 2017 - Planet Friendly Web Development
   @mrchrisadams
   © 2011-2012 GreenTouch Consortium
   2020 ICT Carbon Footprint
   820m tons CO2
   360m tons CO2
   260m tons CO2
   2007 Worldwide ICT

   carbon footprint:

   2% = 830 m tons CO2
   Comparable to the

   global aviation 

   industry
   Expected to grow 

   to 4% by 2020
   The Climate Group, GeSI
   Report “Smart 2020”, 2008
   Full report - https://www.slideshare.net/greentouch-org/intro-to-green-touch
   

   View Slide

10. Pycon Italia 2017 - Planet Friendly Web Development
    @mrchrisadams
    A mental model for
    thinking about this
    

    View Slide

11. Pycon Italia 2017 - Planet Friendly Web Development
    @mrchrisadams
    Your Servers
    Your Packets
    Your Process
    

    View Slide

12. Your servers
    Provider
    Provisioning
    

    View Slide

13. Your servers
    Provider:
    Where did the power your
    servers run on come from?
    

    View Slide

14. Power quote
    From 2017 report - http://www.greenpeace.org/usa/global-warming/click-clean/
    

    View Slide

15. Your servers
    Provision:
    How many server resources do
    you use? And when?
    

    View Slide

16. The power of wireless cloud 2013 (CEET - http://www.ceet.unimelb.edu.au/publications)
    

    View Slide

17. “many workloads are spiky or
    drop to zero for extended
    periods… applications used in
    the workplace may only be
    active for 40 of the 168 hours in
    a week.”
    Adrian Cockroft
    @adrianco: AWS, ex Netflix
    @mrchrisadams
    https://medium.com/@adrianco/evolution-of-business-logic-from-monoliths-through-microservices-to-functions-
    ff464b95a44d#.4ap421dwr
    

    View Slide

18. PaaS:
    Higher unit cost in
    exchange for
    easier scaling
    

    View Slide

19. FaaS / Serverless:
    Even higher unit
    cost, to not think
    about scaling*
    * You still need to obviously, but differently
    

    View Slide

20. “the technical capabilities of
    serverless might not be that
    exciting, but the financial side
    sure is… Strong financial
    incentives for good design
    practices, and clear financial
    penalties for bad design”
    Gojko Adzic
    @mrchrisadamshttps://gojko.net/2016/08/27/serverless.html
    

    View Slide

21. When your second biggest cost after
    staff is data centres, you have a very
    good incentive to care about saving
    energy.
    @mrchrisadams
    https://deepmind.com/blog/deepmind-ai-reduces-google-data-centre-cooling-bill-40
    

    View Slide

22. Platform as a Service (PaaS) Functions as a Service
    Cloudfoundry: App Autoscaler chalice
    Heroku: Autoscaling & LightSwitch zappa
    In the Python toolbox
    

    View Slide

23. Your packets
    Lossy changes
    Lossless changes
    

    View Slide

24. http://www.electricitymap.org/
    

    View Slide

25. “Wireless access network
    technologies account for 90% of
    total wireless cloud energy
    consumption. Data centres account
    for only about 9%. The energy
    consumption of wireless user
    devices is negligible.”
    The power of wireless cloud 2013 (CEET - http://www.ceet.unimelb.edu.au/publications)
    

    View Slide

26. Power quote
    The power of wireless cloud 2013 (CEET - http://www.ceet.unimelb.edu.au/publications)
    

    View Slide

27. https://www.soasta.com/blog/page-bloat-2015-web-performance-monitoring/
    

    View Slide

28. Power quote
    https://performancebudget.io
    

    View Slide

29. Power quote
    https://speedcurve.com (see also sitespeed.io for an OSS take)
    

    View Slide

30. Your packets
    Lossy changes:
    How much of what you send is
    used by the client?
    

    View Slide

31. Your packets
    Lossless changes:
    How efficiently are the bits
    sent?
    

    View Slide

32. Dynamic Static Media
    compression
    Django middleware,
    Flask-Compress, etc.
    Whitenoise, CDNs,
    Nginx
    CDNs, Nginx
    caching
    Django middleware &
    decorators, Flask-Cache
    Whitenoise, CDNs CDNs, Nginx
    JS optimisation
    Webpack 2, Google
    Closure Compiler (JS)
    nginx mod_pagespeed
    Image Optimisation ImageOptim, Trimage
    easy-
    thumbnails.optimise,
    nginx mod_pagespeed
    Relevant in the Python toolbox
    

    View Slide

33. There’s much, much more to do, but
    starting here can help
    

    View Slide

34. Thanks!
    I’m around all week, and in the
    slack channel to chat. Say
    hello :)
    Chris Adams
    @mrchrisadams
    planetfriendly.productscience.co.uk
    speakerdeck.com/mrchrisadams
    

    View Slide

35. Pycon Italia 2017 - Planet Friendly Web Development
    @mrchrisadams
    Bonus life-changing link:
    What can a technologist do about
    climate change? - Bret Viktor
    worrydream.com/climatechange
    

    View Slide