These Five Tricks Can Make Your Apps Greener, Cheaper, & Nicer

The code we write has a climate impact. But how big is that impact? How do we measure it? How do we reduce it? Is the
cloud helping? What’s going on with Virginia? Are we still allowed to do CI/CD? Will native compilation save us? Is Java
even a good choice anymore? This talk discusses some of the trade-offs for a modern software developer, and provides a
roadmap to figuring out the right thing.

Transcript

1. Holly Cummins Senior Principal Software Engineer, Quarkus ADEO x Decathlon Digital Dev Summit May 27, 2026 these five tricks can make your apps greener cheaper & nicer
2. @holly_cummins #RedHat Sources: https://www.greenit.fr/wp-content/uploads/2019/11/GREENIT_EENM_etude_EN_accessible.pdf https://ourworldindata.org/ghg-emissions-by-sector the digital world creates more carbon emissions than aviation
3. @holly_cummins #RedHat Lighter area represents high and low estimates, where available. Sources: https://www.iea.org/fuels-and-technologies/data-centres-networks https://ourworldindata.org/grapher/electricity-demand?tab=table&country=USA~GBR~FRA~DEU~IND~BRA data centres use as much electricity as a medium country
4. @holly_cummins #RedHat it’s not just artificial intelligence
5. @holly_cummins #RedHat it’s not just artificial intelligence it’s not just cryptocurrency mining
6. @holly_cummins #RedHat it’s not just artificial intelligence it’s not just cryptocurrency mining it’s all of us
7. @holly_cummins #RedHat aaaaaaaargh
8. @holly_cummins #RedHat aaaaaaaargh?
9. @holly_cummins #RedHat we have solutions
10. @holly_cummins #RedHat green software foundation: principles
11. @holly_cummins #RedHat carbon awareness green software foundation: principles
12. @holly_cummins #RedHat carbon awareness green software foundation: principles where
13. @holly_cummins #RedHat carbon awareness green software foundation: principles where when
14. @holly_cummins #RedHat carbon awareness green software foundation: principles hardware efficiency where when
15. @holly_cummins #RedHat carbon awareness green software foundation: principles hardware efficiency where when elasticity
16. @holly_cummins #RedHat carbon awareness green software foundation: principles hardware efficiency where when elasticity utilisation
17. @holly_cummins #RedHat carbon awareness green software foundation: principles hardware efficiency electricity efficiency where when elasticity utilisation
18. @holly_cummins #RedHat algorithms carbon awareness green software foundation: principles hardware efficiency electricity efficiency where when elasticity utilisation
19. @holly_cummins #RedHat algorithms stack carbon awareness green software foundation: principles hardware efficiency electricity efficiency where when elasticity utilisation
20. @holly_cummins #RedHat algorithms stack carbon awareness hardware efficiency electricity efficiency where when elasticity utilisation green software foundation: principles
21. @holly_cummins #RedHat what programming languages create the least carbon? stack
23. @holly_cummins #RedHat
24. energy consumption of programming languages
25. @holly_cummins rust
26. @holly_cummins rust rust is hard
27. @holly_cummins rust trading off human efficiency against machine efficiency rust is hard
28. @holly_cummins java is nice
29. @holly_cummins #RedHat not all java is equally efficient
30. digression: measuring carbon is hard
31. @holly_cummins #RedHat step 1: measure power usage
32. @holly_cummins #RedHat step 1: measure power usage wall power measurement
33. @holly_cummins #RedHat step 1: measure power usage wall power measurement more complete
34. @holly_cummins #RedHat step 1: measure power usage wall power measurement more complete needs access to the wall
35. @holly_cummins #RedHat step 1: measure power usage wall power measurement more complete needs access to the wall RAPL
36. @holly_cummins #RedHat step 1: measure power usage wall power measurement more complete needs access to the wall RAPL programmatically accessible
37. @holly_cummins #RedHat step 1: measure power usage wall power measurement more complete needs access to the wall RAPL programmatically accessible misses some components
38. @holly_cummins #RedHat step 1: measure power usage wall power measurement more complete needs access to the wall data costs carbon RAPL programmatically accessible misses some components
39. @holly_cummins #RedHat
40. @holly_cummins #RedHat
41. @holly_cummins #RedHat
42. @holly_cummins #RedHat load
43. @holly_cummins #RedHat Source: Teads EC2 instances carbon dataset
44. @holly_cummins #RedHat coal wind step 2: convert power usage to carbon solar
45. @holly_cummins #RedHat published energy mixes (these are made-up energy mixes)
46. @holly_cummins #RedHat step 3: embedded carbon (manufacturing has costs)
47. @holly_cummins #RedHat simpler models?
48. @holly_cummins #RedHat all models are wrong, some are useful
49. @holly_cummins #RedHat
50. @holly_cummins #RedHat these two columns are almost the same
51. @holly_cummins #RedHat Energy 1 10 100 Time 1 10 100 energy efficiency across programming languages Python Rust Java Go
52. @holly_cummins #RedHat Energy 1 10 100 Time 1 10 100 the trend line is more or less straight energy efficiency across programming languages Python Rust Java Go
53. @holly_cummins #RedHat energy consumption (sort of, mostly) is proportional to execution time* * assuming hardware and other resource stays the same
54. @holly_cummins #RedHat trick 1: the vrrrrrooooooooom model* * a made-up name
55. @holly_cummins #RedHat trick 1: the vrrrrrooooooooom model* * a made-up name faster software is greener software
56. @holly_cummins
57. @holly_cummins invented by Dr. Vroom (really!)
58. @holly_cummins naming is the hardest problem in computer science invented by Dr. Vroom (really!)
59. @holly_cummins naming is the hardest problem in computer science
60. @holly_cummins
61. @holly_cummins #RedHat not all java is equally fast
62. @holly_cummins #RedHat not all java is equally efficient
63. case study quarkus
64. @holly_cummins #RedHat
65. @holly_cummins #RedHat
66. @holly_cummins #RedHat does being small and fast reduce carbon footprint?
67. #Quarkus @holly_cummins measure, don’t guess.
68. did measurements show quarkus is more efficient?
69. @holly_cummins #RedHat Source: Clement Escoffier cost impact of framework choice Setup: • 800 requests/second, over 20 days • SLA > 99% • AWS instances Assumptions: • Costs are for us-east-1 data centre
70. @holly_cummins #RedHat
71. @holly_cummins #RedHat to measure energy, we would need:
72. @holly_cummins #RedHat to measure energy, we would need: (a) hardware access
73. @holly_cummins #RedHat to measure energy, we would need: (a) hardware access (b) time machine
74. @holly_cummins #RedHat load to measure energy, we would need: (a) hardware access (b) time machine
75. @holly_cummins #RedHat Source: Teads EC2 instances carbon dataset
76. @holly_cummins #RedHat Source: Clement Escoffier cost impact of framework choice Setup: • 800 requests/second, over 20 days • SLA > 99% • AWS instances Assumptions: • Costs are for us-east-1 data centre
77. @holly_cummins #RedHat cost and carbon impact of framework choice Carbon estimated using TEADS dataset
78. @holly_cummins #RedHat trick 2: the economic model*
79. @holly_cummins #RedHat trick 2: the economic model* * "economic input-output life cycle assessment"
80. @holly_cummins #RedHat reducing your cloud spend (probably) reducing your carbon footprint* hardware spend electricity bill * if you keep other factors the same
81. @holly_cummins #RedHat Setup: • 800 requests/second, over 20 days • SLA > 99% Assumptions: • 50% load • us-east-1 data centre • Teads dataset Source: Clement Escoffier x Teads cloud carbon impact of framework choice
82. @holly_cummins #RedHat Setup: • 800 requests/second, over 20 days • SLA > 99% Assumptions: • 50% load • us-east-1 data centre • Teads dataset Source: Clement Escoffier x Teads cloud carbon impact of framework choice economic model in action: the cost and carbon metrics are (roughly) the same
83. @holly_cummins #RedHat what if we measure on-prem?
84. @holly_cummins #RedHat Setup: • REST + CRUD • large heap • RAPL energy measurement • multiple instances to support high load Assumptions: • US energy mix Source: John O’Hara carbon as a function of load
85. @holly_cummins #RedHat Setup: • REST + CRUD • large heap • RAPL energy measurement • multiple instances to support high load Assumptions: • US energy mix Source: John O’Hara carbon as a function of load wait, what?
86. @holly_cummins #RedHat Setup: • REST + CRUD • large heap • RAPL energy measurement • multiple instances to support high load Assumptions: • US energy mix Source: John O’Hara carbon as a function of load
87. @holly_cummins #RedHat capacity Source: John O’Hara Setup: • REST + CRUD • large heap • RAPL energy measurement Assumptions: • US energy mix carbon as a function of load (single instance)
88. @holly_cummins #RedHat capacity Source: John O’Hara Setup: • REST + CRUD • large heap • RAPL energy measurement Assumptions: • US energy mix carbon as a function of load (single instance) shorter line means lower max throughput
89. @holly_cummins #RedHat capacity Source: John O’Hara Setup: • REST + CRUD • large heap • RAPL energy measurement Assumptions: • US energy mix carbon as a function of load (single instance) shorter line means lower max throughput higher line means worse carbon footprint
90. @holly_cummins #RedHat capacity Source: John O’Hara Setup: • REST + CRUD • large heap • RAPL energy measurement Assumptions: • US energy mix carbon as a function of load (single instance) vrrrooom model in action: quarkus on JVM has the smallest footprint … because it has the highest throughput shorter line means lower max throughput higher line means worse carbon footprint
91. @holly_cummins #RedHat trick 3: use quarkus
92. @holly_cummins #RedHat trick 3: use quarkus quarkus ‘automatically’ saves
93. @holly_cummins #RedHat trick 3: use quarkus quarkus ‘automatically’ saves • time
94. @holly_cummins #RedHat trick 3: use quarkus quarkus ‘automatically’ saves • time • money
95. @holly_cummins #RedHat trick 3: use quarkus quarkus ‘automatically’ saves • time • money • carbon (~2x)
96. @holly_cummins #RedHat trick 3: use quarkus quarkus ‘automatically’ saves • time • money • carbon (~2x) • … even when Spring compatibility libraries are used (almost no code changes except dependencies and tests)
97. @holly_cummins #RedHat trick 3: use quarkus quarkus ‘automatically’ saves • time • money • carbon (~2x) • … even when Spring compatibility libraries are used (almost no code changes except dependencies and tests)
98. @holly_cummins #RedHat low load choose quarkus native normal load choose quarkus on jvm
99. @holly_cummins #RedHat carbon awareness green software foundation: principles hardware efficiency electricity efficiency
100. @holly_cummins #RedHat carbon awareness green software foundation: principles hardware efficiency electricity efficiency elasticity
101. @holly_cummins #RedHat carbon awareness green software foundation: principles hardware efficiency electricity efficiency elasticity utilisation
102. @holly_cummins #RedHat application utilisation
103. @holly_cummins #RedHat application utilisation high utilisation good case
104. @holly_cummins #RedHat application utilisation over-utilisation very bad case
105. @holly_cummins #RedHat application utilisation over-utilisation very bad case under-utilisation wasteful case
106. @holly_cummins #RedHat application elasticity high utilisation good case @holly_cummins
107. @holly_cummins #RedHat application elasticity scale-up good utilisation @holly_cummins
108. @holly_cummins #RedHat application elasticity scale-down good utilisation @holly_cummins
109. @holly_cummins #RedHat native quarkus is great at elasticity: starts faster than a light bulb
110. @holly_cummins #RedHat … but elasticity isn’t just a technical problem
111. #RedHat @hollycummins.com Hey boss, I created a Kubernetes cluster. 2018
112. #RedHat @hollycummins.com Hey boss, I created a Kubernetes cluster. I forgot it for 2 months. 2018
113. #RedHat @hollycummins.com Hey boss, I created a Kubernetes cluster. I forgot it for 2 months. … and it’s €1000 a month. 2018
114. #RedHat @hollycummins.com
115. #RedHat @holly_cummins 2017 server-survey 25% doing no useful work (16,000 sampled)
116. #RedHat @holly_cummins 2017 server-survey 25% doing no useful work (16,000 sampled) “perhaps someone forgot to turn them off”
117. #RedHat @holly_cummins@hachyderm.io https://www.anthesisgroup.com/wp-content/uploads/2019/11/Comatose-Servers-Redux-2017.pdf 2014 server-survey 29% active less than 5% of the time (4,000 sampled)
118. @holly_cummins #RedHat cloud elasticity? https://www.business2community.com/cloud-computing/overprovisioning-always-on-resources-lead-to-26-6-billion-in-public-cloud-waste-expected-in-2021-02381033
119. @holly_cummins #RedHat cloud elasticity? 2021: https://www.business2community.com/cloud-computing/overprovisioning-always-on-resources-lead-to-26-6-billion-in-public-cloud-waste-expected-in-2021-02381033
120. @holly_cummins #RedHat cloud elasticity? 2021: $26.6 billion wasted https://www.business2community.com/cloud-computing/overprovisioning-always-on-resources-lead-to-26-6-billion-in-public-cloud-waste-expected-in-2021-02381033
121. @holly_cummins #RedHat cloud elasticity? 2021: $26.6 billion wasted by always-on cloud instances https://www.business2community.com/cloud-computing/overprovisioning-always-on-resources-lead-to-26-6-billion-in-public-cloud-waste-expected-in-2021-02381033
122. @holly_cummins #RedHat it’s not just money
123. @holly_cummins #RedHat it’s not just electricity it’s not just money
124. @holly_cummins #RedHat it’s not just electricity it’s water it’s not just money
125. @holly_cummins #RedHat it’s not just electricity it’s water it’s e-waste it’s not just money
126. #RedHat @hollycummins.com shutting systems off can seem scary
127. @hollycummins.com #RedHat ultimate elasticity
128. #RedHat @hollycummins.com we don’t switch the light off because we’re not sure if it will come back on never happens
129. #RedHat @hollycummins.com we don’t switch the server off because we’re not sure if it will come back on happens all the time
130. #RedHat @hollycummins.com we don’t switch the light off because it’s so much effort to turn it back on never happens
131. #RedHat @hollycummins.com we don’t switch the server off because it would be too much work to recreate it happens all the time
132. #RedHat @hollycummins.com we don’t switch the server off because we’re not really sure if it we’re using it happens all the time
133. @holly_cummins #RedHat what if … turning applications off was no more scary than turning the lights off? ultimate elasticity
134. @holly_cummins #RedHat
135. @holly_cummins #RedHat
136. @holly_cummins #RedHat turning it off and on again must
137. @holly_cummins #RedHat turning it off and on again must • be fast
138. @holly_cummins #RedHat turning it off and on again must • be fast • actually work
139. @holly_cummins #RedHat turning it off and on again must • be fast • actually work • idempotency
140. @holly_cummins #RedHat turning it off and on again must • be fast • actually work • idempotency • resiliency
141. @holly_cummins #RedHat architect things to be turned off and on often
142. @holly_cummins #RedHat trick 4: LightSwitchOps architect things to be turned off and on often
143. @holly_cummins #RedHat trick 4: LightSwitchOps architect things to be turned off and on often
144. @hollycummins.com #RedHat step 1: no more scary state
145. #RedHat @hollycummins.com GitOps (infrastructure as code)
146. #RedHat @hollycummins.com GitOps (infrastructure as code)
147. #RedHat @hollycummins.com
148. #RedHat @hollycummins.com spin it down
149. #RedHat @hollycummins.com kubectl apply -f all-my-cluster/ spin it down spin it up
150. #RedHat @hollycummins.com kubectl apply -f all-my-cluster/ spin it down spin it up
151. #RedHat @hollycummins.com kubectl apply -f all-my-cluster/ ansible-playbook stuff.yml spin it down spin it up
152. @hollycummins.com #RedHat step 1: no more scary state step 2: automate, automate
153. #RedHat @hollycummins.com large UK bank, 2013 50% reduction in CPUs with a lease system self-destructing instances
154. @holly_cummins #RedHat scripting we used to leave our applications running all the time when we scripted turning them off at night, we reduced our cloud bill by 30% @darkandnerdy, Chicago DevOpsDays
155. @holly_cummins #RedHat
156. @holly_cummins #RedHat
157. @holly_cummins #RedHat
158. @holly_cummins #RedHat quarkus!
159. #RedHat @hollycummins.com all the —opses
160. #RedHat @hollycummins.com GreenOps
161. #RedHat @hollycummins.com GreenOps greenops is a mid-sized trilobite (really)
162. @holly_cummins #RedHat FinOps figuring out who in your company forgot to turn off their cloud
163. #Quarkus @holly_cummins you can’t optimise what you can’t measure
164. @holly_cummins #RedHat
165. @holly_cummins #RedHat backstage.io
166. @holly_cummins #RedHat backstage.io •cost insights plugin
167. @holly_cummins #RedHat backstage.io •cost insights plugin •cloud carbon footprint plugin
168. @holly_cummins #RedHat algorithms stack carbon awareness hardware efficiency electricity efficiency where when elasticity utilisation green software foundation: principles
169. trick 5: electricity source
170. not all energy is equal
171. @holly_cummins #RedHat data center location matters
172. @holly_cummins #RedHat
173. @holly_cummins #RedHat
174. @holly_cummins #RedHat we need to talk about virginia
175. @holly_cummins #RedHat look at the sustainability information before choosing a hosting region
176. @holly_cummins #RedHat look at the sustainability information before choosing a hosting region choose a cloud provider who make this easy
177. @holly_cummins #RedHat time of day matters • (most) renewables are intermittent • if grid load is high, shortfalls are filled by fossil fuels
178. @holly_cummins #RedHat what about the ai-lephant in the room?
179. @holly_cummins #RedHat
180. @holly_cummins #RedHat tokenmaxxing
181. @holly_cummins #RedHat ai trick i do not add “destroy planet” to your targets
182. @holly_cummins #RedHat ai trick i do not add “destroy planet” to your targets
183. @holly_cummins #RedHat ai trick ii don’t use a bigger model than you need
184. @holly_cummins #RedHat remember the vrrrrrooooooooom model?
185. @holly_cummins #RedHat remember the vrrrrrooooooooom model? faster software is greener software
186. @holly_cummins #RedHat bigger models are slower
187. @holly_cummins #RedHat 1.3 - 2.2x faster https://www.reddit.com/r/ClaudeAI/comments/1r9jf2j/i_benchmarked_opus_46_vs_sonnet_46_on_agentic_pr/ claude sonnet is than opus
188. @holly_cummins #RedHat ai trick ii don’t use a more expensive model than you need
189. @holly_cummins #RedHat ai trick ii don’t use a more expensive model than you need
190. @holly_cummins #RedHat reminder: economic model
191. @holly_cummins #RedHat reducing your token spend (probably) reducing your carbon footprint* subscription level * if you keep other factors the same
192. @holly_cummins #RedHat 1.7 - 5.5x costlier https://www.reddit.com/r/ClaudeAI/comments/1r9jf2j/i_benchmarked_opus_46_vs_sonnet_46_on_agentic_pr/ claude sonnet is than opus
193. @holly_cummins #RedHat 10 - 50x cheaper https://www.reddit.com/r/ClaudeAI/comments/1r9jf2j/i_benchmarked_opus_46_vs_sonnet_46_on_agentic_pr/ deep seek is for a small* quality drop
194. @holly_cummins #RedHat ai trick iii don’t route trivial commands through an LLM
195. @holly_cummins #RedHat
196. @holly_cummins #RedHat
197. @holly_cummins #RedHat
198. @holly_cummins #RedHat ai trick iv offload to deterministic tools
199. @holly_cummins #RedHat 1. Create the project using quarkus_create with the extensions above. This will auto-start dev mode with a PostgreSQL Dev Service (no manual DB setup needed). │ │ 2. Load extension skills via quarkus_skills to learn the correct patterns for REST and Panache. │ │ 3. Verify the app starts successfully and the generated starter endpoints work by checking the dev mode logs. │ │ │ │ Verification │ │ │ │ - Confirm dev mode starts without errors │ │ - Confirm PostgreSQL Dev Service is running (auto-provisioned by Quarkus) │ │ - Test the generated starter endpoint(s) via curl or the Dev UI ❯ /usage Total duration (API): 58s Total duration (wall): 2m 39s Total code changes: 34 lines added, 0 lines removed Usage by model: claude-opus-4-6: 2.3k input, 2.6k output, 160.7k cache read, 44.7k cache write ($0.4364) 10x generating a quarkus app cheaper with quarkus mcp
200. @holly_cummins #RedHat like using a lamborghini to mow the lawn
201. @holly_cummins #RedHat ai trick iv time of day matters location matters
202. @holly_cummins #RedHat ai trick iv time of day matters location matters
203. @holly_cummins #RedHat aaaaaaaargh?
204. @holly_cummins #RedHat efficient software:
205. @holly_cummins #RedHat software that does less of what you don’t need efficient software:
206. @holly_cummins #RedHat efficient software:
207. @holly_cummins #RedHat software that does what you need using fewer resources efficient software:
208. @holly_cummins #RedHat “no-regrets” solutions
209. @holly_cummins #RedHat
210. @holly_cummins #RedHat co-benefits
211. @holly_cummins #RedHat co-benefits the double win
212. @holly_cummins #RedHat co-benefits the double win win-win
213. @holly_cummins #RedHat co-benefits the double win win-win twofer
214. @holly_cummins #RedHat co-benefits the double win win-win twofer überwinden
215. @holly_cummins #RedHat co-benefits the double win win-win twofer überwinden
216. @holly_cummins #RedHat climate solutions can make everything better
217. @holly_cummins #RedHat the double-win turning things off saves a lot of money
218. @holly_cummins #RedHat the double-win solar power can be really inexpensive
219. @holly_cummins #RedHat remember the zombie servers? https://www.business2community.com/cloud-computing/overprovisioning-always-on-resources-lead-to-26-6-billion-in-public-cloud-waste-expected-in-2021-02381033
220. @holly_cummins #RedHat remember the zombie servers? what else could we do with that $26.6 billion of wasted cloud spend? https://www.business2community.com/cloud-computing/overprovisioning-always-on-resources-lead-to-26-6-billion-in-public-cloud-waste-expected-in-2021-02381033
221. @holly_cummins #RedHat remember the vrrrrrooooooooom model? (probably not, it was a made-up name)
222. @holly_cummins #RedHat the double-win
223. @holly_cummins #RedHat the double-win
224. @holly_cummins #RedHat “this is not sacrifice. it’s advancement.” – Dr. Jonathan Foley
225. @holly_cummins #RedHat tl;dpa (too long; didn’t pay attention) trick 1: the vrrrooooom model says faster is greener trick 2: the economic model says cheaper is greener trick 3: choose a fast and energy-efficient framework (such as quarkus) trick 4: architect to be able to turn stuff off (LightSwitchOps) trick 5: choose your time of day and hosting wisely
226. @holly_cummins #RedHat we all make a difference
227. slides thank you Holly Cummins Red Hat