Paul Johnston is an expert in serverless technologies and was formerly the CTO of a serverless startup. In the document, he discusses whether serverless is truly the future and compares it to container-based architectures. While containers can allow for increased deployment efficiency and scale, serverless architectures require less maintenance overhead and management burden. Serverless solutions are also more cost-effective since the provider manages the underlying infrastructure. Overall, Johnston argues that serverless will be the future for most use cases due to its automation, high availability, reduced complexity, and increased feature velocity compared to managing one's own servers and containers.
1. Serverless is the future.
(Or is it?)
Paul Johnston @PaulDJohnston Medium/Twitter
2. Paul Johnston
Twitter and Medium: @PaulDJohnston
Discuss Serverless / Being a CTO /
Climate Change
Interim CTO/Consultant (currently
available)
Ex-AWS Senior Developer Advocate for
Serverless
Co-founder JeffConf and ServerlessDays
Former CTO of Movivo - one of the
world’s first Serverless startups
4. Future of Cloud and Climate
Data Centres currently 2% of Global Carbon Emissions (at least)
Growth of Cloud/Data Centres likely to grow by at least 5x in next 7 years
Efficiency is irrelevant due to Jevons Paradox - increased efficiency leads to
increased demand)
Not just climate, but energy security - energy price rises
Whitepaper written by Anne Currie and myself (to be released soon)
5. ETHICS WHITEPAPER - THE STATE OF DATA
CENTRE ENERGY USE IN 2018
bit.ly/2024wp
8. Movivo
Became CTO in 2015
Android App + AWS Lambda
By early 2017 was in 20+ countries with
over 500,000 MAU
AWS bill was around $300/month (half
was data backup)
Team of 2 serverless and 2 android
developers in total
9. Movivo
Heavily used Terraform
Lots of managed services
☹ 1 instance for fixed IP proxy to
Lambda
100+ functions deployed
No RDBMS - Primarily DynamoDB and
Athena
Dev envy + Staging + Prod + CI +
GitHub
10. So why didn’t I use
Containers?
(it was 2015 after all)
74. Serverless beats Containers
• You don’t have to maintain the containers as well as the code (feature
velocity increases)
• Providers manage maintenance (security win)
• Efficiency is increased when you only have to manage one aspect
81. For the 99% Serverless Wins
• Serverless scales for most scenarios really well
• If you’re that big (the 1%) then you can probably build a big team so do
containers
• Even then why would you?
82. Oh! One thing…
• Serverless is built on Containers
• Containers are still important
• Managed Services will negate Containers
• We will always need Containers… for the 1%
86. Serverless Thinking
• Use a service whenever possible (because services are cheaper than people)
• Local development is much less important (or useful)
• Constraints are not bad (e.g. FaaS python version, memory and cores)
• It’s easier with Greenfield but can have real value in replacing elements of legacy
applications
• Aim is to avoid running your own servers as much as possible
• Vendors mostly have better engineers than you do
• It’s an infrastructure solution not an application solution