How Building Greener Software Cuts Your Bottom Line

How Building Greener Software Cuts Your Bottom Line ![rw-book-cover](https://img.youtube.com/vi/WsZCn8eXsuY/maxresdefault.jpg) 🔗 **Source:** [How Building Greener Software Cuts Your Bottom Line](https://www.youtube.com/watch?v=WsZCn8eXsuY) > [!note]+ Document Notes > tags: #green-software #finops #cloud-optimization #digital-sustainability #carbon-footprint #cloud-cost > author: Lead Dev > source: YouTube > reference: https://www.youtube.com/watch?v=WsZCn8eXsuY > date created: 2024-12-19 14:30 > > # 🏷️ Title: How building greener software cuts your bottom line >  ## 🧰 Action Inbox - [H1] Implement shift-left sustainability decisions in development standards — avoid 50% waste before deployment #green-software - [H2] Integrate environmental metrics into existing FinOps dashboards — prevent dashboard fatigue #finops - [H3] Challenge leadership budgeting practices that encourage resource bloat — address root cause waste #cloud-optimization ## 🧭 Table of Contents - [[#📝 Summary]] - [[#📖 Main Content]] - [[#🗒️ Notes]] - [[#📚 References]] - [[#🗣️ Paraphrase]] - [[#📰 Newsletter]] - [[#✅ Action Items]] - [[#🧪 Hypotheses]] - [[#📊 Methodology]] - [[#❓ Questions]] - [[#🔮 Future Research]] - [[#💡 Implications]] - [[#🧠 Learning Styles Assessment]] - [[#📊 Accessibility Assessment]] - [[#🌍 DEIB Assessment]] - [[#✅ Logical Validity]] - [[#🛡️ DARVO Analysis]] - [[#🎯 Conclusion]] ### 🔗 References - [Lead Dev Community](https://slack.theleaddev.com) - [Harness Platform](https://harness.io) ## 📝 Summary - Organizations waste approximately 50% of their IT budget on inefficient cloud resources - GreenOps aligns environmental sustainability with cost optimization through operational improvements - Leadership buy-in requires connecting sustainability metrics to existing business priorities like security and compliance - "Shift-left" approach prevents waste at design phase rather than optimizing after deployment - Environmental metrics can motivate teams when cost savings alone fail to drive change ## 📖 Main Content This panel discussion explores how environmental sustainability practices ("GreenOps") can dramatically reduce cloud costs while improving operational efficiency. The speakers argue that traditional cloud marketing overstates efficiency benefits, while poor budgeting practices and organizational silos create massive waste. FinOps (Finance + DevOps) focuses on deriving business value from every cloud dollar spent, while GreenOps applies sustainability principles across all IT operations. These approaches align directionally - right-sizing instances, scheduling resources, and eliminating zombie services benefit both cost and carbon footprint. The panelists emphasize that successful implementation requires cultural change rather than new tools. Environmental impact provides an additional lens for engaging leadership who may not respond to cost arguments alone. However, the most effective approach integrates sustainability metrics into existing workflows rather than creating separate reporting systems. ## 🗒️ Notes #### ⏱️ [00:00](https://youtu.be/WsZCn8eXsuY?t=0) — [H1] Introduction and panel setup - Key idea: Panel explores how greener software can cut cloud costs by ~50% - Evidence: "Building greener software which oh by the way can cut your cloud cost by approximately 50%" #### ⏱️ [02:30](https://youtu.be/WsZCn8eXsuY?t=150) — [H2] Defining FinOps vs GreenOps - Key idea: FinOps is Finance + DevOps focused on business value from cloud spend - Evidence: "FinOps definitely an interesting portmanteau... it is a mashup of finance and DevOps" #### ⏱️ [04:45](https://youtu.be/WsZCn8eXsuY?t=285) — [H1] GreenOps covers entire IT operations, not just cloud - Key idea: GreenOps applies sustainability principles to all IT infrastructure and processes - Evidence: "Green ops is the principle that covers the entire operations of it. It's not just cloud. It's data centers, infrastructure, software development" #### ⏱️ [07:20](https://youtu.be/WsZCn8eXsuY?t=440) — [H2] Cloud provider sustainability marketing is misleading - Key idea: Cloud providers cherry-pick best-case scenarios against worst-case on-premise setups - Evidence: "They pick the most practically perfect cloud deployment that's never existed and they balance that against the ugliest most horrible infrastructure" #### ⏱️ [10:15](https://youtu.be/WsZCn8eXsuY?t=615) — [H3] FinOps and GreenOps don't always align - Key idea: Some FinOps practices like spend-based commitments discourage efficient consumption - Evidence: "Spend-based commitments... does not encourage efficient consumption because there's no motivation to turn something off below a threshold" #### ⏱️ [13:30](https://youtu.be/WsZCn8eXsuY?t=810) — [H1] Typical enterprise waste reaches 50% of IT spend - Key idea: Organizations waste about 50 cents of every IT dollar through overdesign and poor practices - Evidence: "Typically we find... it's normally about 50 cents in every dollar is wasted on IT" #### ⏱️ [16:45](https://youtu.be/WsZCn8eXsuY?t=1005) — [H1] Environmental framing helps win leadership support - Key idea: Sustainability messaging reaches executives who aren't motivated by cost savings alone - Evidence: "Leadership aren't actually motivated that much by cost savings... if you take the environmental lens to this you catch enough of them" #### ⏱️ [19:20](https://youtu.be/WsZCn8eXsuY?t=1160) — [H1] Shift-left approach prevents waste at design phase - Key idea: Making sustainable decisions during ideation is far easier than retrofitting live systems - Evidence: "If eight months before when this was in the ideation phase, I came to you... Those decisions aren't baked in yet" #### ⏱️ [23:10](https://youtu.be/WsZCn8eXsuY?t=1390) — [H2] Budgeting practices encourage massive resource bloat - Key idea: Separate run/change budgets force teams to over-provision resources upfront - Evidence: "They encourage massive oversizing of resources, massive overcommitment because they don't know what they might need in subsequent years" #### ⏱️ [26:40](https://youtu.be/WsZCn8eXsuY?t=1600) — [H2] Integration beats new dashboards for sustainability data - Key idea: Embed environmental metrics in existing tools rather than creating separate systems - Evidence: "Get that data into the tools and systems that you're already working in... get it into your FinOps view" #### ⏱️ [30:15](https://youtu.be/WsZCn8eXsuY?t=1815) — [H3] Multiple personas need different metrics and visualization - Key idea: No single KPI works for all stakeholders - tailor metrics to each audience - Evidence: "There is no one KPI to rule them all... everyone looks at this problem through a different lens" #### ⏱️ [35:20](https://youtu.be/WsZCn8eXsuY?t=2120) — [H2] AI workloads driving cloud-to-datacenter migration - Key idea: AI costs in cloud are so high that organizations return to on-premise infrastructure - Evidence: "If you oversize something on an AI platform, it's not a million dollar mistake, it's a $30 million mistake" #### ⏱️ [38:45](https://youtu.be/WsZCn8eXsuY?t=2325) — [H3] Water usage as more tangible environmental metric than carbon - Key idea: Concrete impacts like water consumption resonate better than abstract carbon numbers - Evidence: "AI data centers are expected to consume six times more fresh water than the entire country of Denmark" #### ⏱️ [42:30](https://youtu.be/WsZCn8eXsuY?t=2550) — [H1] First step recommendations for GreenOps adoption - Key idea: Start with existing optimization practices while adding environmental visibility - Evidence: "Don't do it in isolation... work with the FinOps teams, work with the engineering teams, bring the visibility of the data into that stream" ## 📚 References - [Lead Dev Slack Community](https://slack.theleaddev.com) - [Harness Cloud Cost Management](https://harness.io) - [FinOps Foundation](https://finops.org) ## 🗣️ Paraphrase - Environmental sustainability and cost optimization align in most cases - efficient resource usage reduces both waste and carbon footprint - Organizations waste enormous amounts on cloud resources due to poor budgeting practices and lack of accountability at executive levels - Success requires cultural change and leadership engagement rather than new tools or separate sustainability teams ## 📰 Newsletter 🌱 **Green Software = Green Budget** New research shows organizations can cut cloud costs by 50% through sustainable engineering practices! The secret isn't new dashboards - it's integrating environmental metrics into your existing FinOps workflows. Key insight: Make sustainability decisions during development, not after deployment. When teams design with resource efficiency in mind, both carbon footprint and costs plummet. **Quick wins:** Right-size instances, schedule dev environment shutdowns, and challenge budget practices that encourage over-provisioning. Resources: [FinOps Foundation](https://finops.org), [Lead Dev Community](https://slack.theleaddev.com), [Cloud Sustainability Guidelines](https://learn.wordpress.org/tutorial/sustainable-web-development/) ## ✅ Action Items #### DevRel - Developer advocacy: Create workshops linking performance optimization to sustainability metrics - Education: Develop "shift-left sustainability" curriculum for engineering teams - Developer experience: Integrate carbon impact data into existing monitoring dashboards - Developer marketing: Campaign highlighting cost-sustainability alignment in cloud practices - Community management: Facilitate GreenOps discussion forums and knowledge sharing #### OSPO - Policy & Governance: Draft sustainable cloud resource usage policies and procurement guidelines - Software Inventory: Include energy efficiency metrics in SBOM and dependency tracking - Developer Support: Training on right-sizing, scheduling, and efficient cloud architectures - Community Engagement: Fund upstream projects focused on resource efficiency and sustainability - Strategic Alignment: Establish KPIs connecting environmental impact to business value #### Community - Maintainers: Develop sustainability-focused roadmaps and resource efficiency targets - Contributors: Create first-issue programs around performance and resource optimization - Sponsors: Transparent reporting on infrastructure carbon footprint and efficiency improvements - Ecosystem: Cross-project coordination on sustainable development practices and standards #### WordPress - Core: Implement resource efficiency budgets alongside performance budgets - Core Performance: Profile and optimize hosting resource consumption patterns - Hosting: Develop cross-host sustainability metrics and efficiency benchmarks - Community: Create meetup content on sustainable WordPress development practices - Documentation: Add sustainability considerations to development best practices guides ## 🧪 Hypotheses - Organizations implementing GreenOps practices will achieve 30-50% cloud cost reduction within 12 months - Environmental metrics will increase leadership engagement in cost optimization initiatives by 40% - Shift-left sustainability decisions will prevent 3x more waste than post-deployment optimization efforts - Integrated sustainability dashboards will see 5x higher adoption than standalone environmental reporting tools ### 📊 Methodology - Longitudinal study tracking cloud spend and resource utilization before/after GreenOps implementation - Survey leadership engagement levels with cost vs. environmental framing of optimization projects - Comparative analysis of waste prevention in early-stage vs. retrofit sustainability initiatives ### ❓ Questions - How can we quantify the business risk of zombie services beyond just cost and carbon impact? - What governance changes are needed to enable cross-team resource optimization at enterprise scale? - How do we balance sustainability goals with performance requirements in latency-sensitive applications? ### 🔮 Future Research - Development of standardized sustainability metrics for software development lifecycles - Investigation of renewable energy scheduling for elastic workloads - Analysis of organizational budget structures that incentivize vs. discourage resource efficiency ### 💡 Implications - Practice: Engineering teams need sustainability training integrated with existing DevOps practices - Policy: Procurement and budgeting policies require restructuring to prevent resource bloat - Security: Zombie service elimination reduces attack surface while cutting costs and emissions - Education: Curriculum must connect environmental impact to concrete business metrics - Funding: Budget allocation methods directly influence resource efficiency and waste generation ### 🧠 Learning Styles Assessment Reading level estimate: 12th grade #### 👀 Visual learners - Present: Some charts mentioned but video format limits visual aids - Improve: Resource utilization dashboards, carbon vs. cost correlation graphs, waste visualization #### 🔊 Aural learners - Present: Full video discussion with expert dialogue - Improve: Structured audio summary highlighting key statistics and action items #### 🗣️ Verbal learners - Present: Clear definitions of FinOps, GreenOps, and industry terminology - Improve: More concrete analogies for abstract concepts like carbon tons #### 🧩 Kinesthetic learners - Present: Discussion of hands-on optimization practices and tool integration - Improve: Step-by-step implementation checklists and measurement frameworks #### 🧠 Logical learners - Present: Cost-benefit analysis and systematic approach to waste elimination - Improve: Decision trees for sustainability vs. performance tradeoffs #### 🧑‍🤝‍🧑 Social learners - Present: Panel discussion format with multiple expert perspectives - Improve: Community implementation guides and peer learning frameworks #### 🧘‍♀️ Solitary learners - Present: Individual reflection points on organizational practices - Improve: Self-assessment tools for current sustainability maturity #### 🌿 Naturalistic learners - Present: Strong focus on environmental impact and ecosystem thinking - Improve: Broader connections to sustainability lifecycle and planetary boundaries ### 📊 Accessibility Assessment #### Screen reader semantics - Present: Video has spoken content but lacks structured transcript - Improve: Proper heading structure in transcript, speaker identification #### Keyboard navigation - Present: Standard YouTube player controls - Improve: Timestamped navigation links, chapter markers #### Contrast and readability - Present: Video format with standard contrast - Improve: High-contrast transcript option, larger text sizing #### Text resizing and reflow - Present: YouTube standard responsive design - Improve: Dedicated transcript page with reflow support #### Media and captions - Present: Auto-generated captions available - Improve: Professional captions with technical term accuracy, downloadable transcript #### Forms and interactions - Present: Basic video player interactions - Improve: Interactive timestamps, searchable content ### 🌍 DEIB Assessment #### Cultural breadth - Present: US and European perspectives on sustainability priorities - Improve: Global South viewpoints on resource access and environmental justice #### Inclusive language - Present: Gender-neutral technical terminology - Improve: Plain language explanations of complex financial concepts #### Neurodiversity support - Present: Structured panel format with clear topic transitions - Improve: Visual topic indicators, consistent pacing #### Representation - Present: Multiple expert voices but limited diversity visible - Improve: Broader range of industry and cultural perspectives #### Psychological safety - Present: Respectful disagreement and acknowledgment of complexity - Improve: Explicit validation of different organizational constraints ### ✅ Logical Validity #### Strong reasoning - Deductive strength: Clear logical connection between resource efficiency and cost reduction - Inductive strength: Patterns identified from enterprise consulting experience - Causal reasoning: Well-supported links between budgeting practices and waste generation - Analogical strength: Effective comparison of cloud vs. datacenter trade-offs #### Fallacies - Hasty generalization: "50% waste" claim needs more specific scope and evidence - Appeal to authority: Heavy reliance on consulting experience without independent verification #### Tone of voice - Overall tone: Professional but accessible, with appropriate urgency about environmental issues ### 🛡️ DARVO Analysis - Deny: No denial patterns present - panelists acknowledge complexity and multiple valid approaches - Attack: No attacks on accusers - constructive criticism of industry practices - Reverse victim/offender: Not applicable to this technical content - Emotional cues and rhetoric: Appropriate urgency about environmental impact balanced with practical business considerations ## 🎯 Conclusion Green software development isn't just about environmental responsibility - it's a strategic approach to eliminating the 50% waste that plagues most organizations' IT spending. Success requires integrating sustainability metrics into existing FinOps practices rather than creating separate environmental initiatives. The key insight is "shifting left" - making resource efficiency decisions during the design phase rather than optimizing after --- ## Full Transcript [00:12] Hello, hello, hello everyone. A huge [00:16] thank you for joining us today for [00:18] today's panel on building greener [00:21] software which oh by the way can cut up [00:23] cut your cloud cost by approximately [00:26] 50%. This is a lead dev online panel [00:30] discussion created in partnership with [00:32] harness. We're going to last about the [00:34] conversation going to last about 45 [00:35] minutes which for both myself and the [00:38] panelists. Then we will head over to the [00:41] lead dev slack to answer any of your [00:43] questions uh in the hashtag panel Q&A [00:47] channel if [00:50] for I'll drop that in for those live. If [00:53] you have questions live we really [00:55] welcome it. [00:56] This got real serious real fast like [00:59] very contentious in a really fun good [01:01] way uh in the planning call. So we would [01:06] love we all agree that the earth is [01:08] important. We all agree that probably [01:12] folks don't care [01:14] enough. Uh so what does that mean? How [01:18] do we influence leadership [01:20] to care about not only budget but carbon [01:25] budget? Uh there's a lot of things that [01:27] can be done. Uh I really do recommend [01:30] you ask questions if you're joining us [01:31] live. Thank you. And if we don't answer [01:34] them live, it'll be in the Slack [01:35] community, which I'll add in the chat [01:37] now. So let's get started with [01:38] introductions. My name is Jennifer [01:40] Riggins. I'm a tech journalist. And then [01:43] I would ask that each of you introduce [01:45] yourselves. So Mark, why don't you kick [01:47] off? [01:49] >> Oh, well, thank you for having me. So [01:50] I'm I'm Mark Butcher from Positive. and [01:53] everything I do is in the world of [01:54] digital sustainability and green ops. So [01:56] I I come to this conversation with with [01:58] a with a fair bias that you can probably [02:00] guess. [02:02] >> Patrick. [02:03] >> Very good. [02:04] >> Yeah. Uh Patrick Broen from Harness and [02:07] I've spent the last six years or so of [02:10] my career building and running Finoff [02:12] practices primarily at large financial [02:14] institutions in the US. [02:17] uh I've run them, you know, sitting in [02:18] the CFO organization as well as in the [02:20] CTO organization, understand the tension [02:23] between those two sides of the house. Uh [02:26] but certainly understand the importance [02:28] of caring about the cloud bill and what [02:31] you can do to make things more [02:32] efficient. I'm now at Harness as a PHOps [02:34] advisory director, which is to help our [02:38] prospective customers and existing [02:39] customers along their own PHOPS maturity [02:43] journeys, whether they're just getting [02:44] started or sort of mid-stream and [02:46] looking to take the next step. And I'm [02:48] excited to be here today. [02:51] >> Okay, thank you both so much. We [02:52] hopefully we'll have a third panelist in [02:55] An Curry. It's 2026. [02:58] We can get to the moon probably. We're [03:00] trying again, but webinars, live [03:04] streams, meetings, you're on mute, all [03:07] of these things, we haven't figured out [03:10] and no one seems to think [03:11] >> maybe I'll start a podcast, call it [03:13] you're on mute. That that'd be a pretty [03:15] good one. [03:16] >> I bet you someone already has. [03:17] >> I'm sure it's already doing. [03:19] >> Um, if it was a real person or an AI [03:21] podcast, who's to know anymore because [03:23] we have that level of sophistication. [03:25] So I just want to start with some [03:27] definitions because we've mentioned [03:29] green ops, we've mentioned FinOps. [03:32] Let's make sure there's still tech loves [03:36] support mento. Let's try to define what [03:39] those are and what's the difference [03:40] since you are director of PHOPS [03:42] advisory. What does that really mean? [03:46] And then I want Mark to tie it together [03:48] if we can in the book. [03:49] >> Sure. [03:50] >> Yeah, I'd love to. So yeah um PHOps [03:53] definitely an interesting portto people [03:55] often think it is short for financial [03:57] operations. [03:59] In fact it is a a mashup of finance and [04:03] DevOps. uh and it operates in much the [04:06] same way that DevOps does where you have [04:08] a number of different personas all sort [04:10] of circled together around the same [04:13] challenge or problem uh and working [04:15] together within their own [04:16] responsibilities [04:18] to in the case of PHOPS ensure that you [04:21] know you're not just aware of your cloud [04:24] spend you're not just trying to reduce [04:26] your cloud spend although that is a [04:28] worthy goal but you're looking to drive [04:31] business value from every dollar that [04:33] you spend on the cloud. So how do you [04:35] tie the expense of a cloud bill whether [04:39] you know it's a steady state and you're [04:41] spending you know the same amount of [04:42] money every year or you're growing your [04:45] cloud spend. How do you derive and [04:48] demonstrate to leadership and company [04:50] executives that the money that they're [04:53] spending on the cloud actually delivers [04:55] business value? So that to me is is [04:57] really the crux of what synopsis is here [04:59] to solve. [05:01] Okay, Mark. Green Ops, which may be a [05:04] made up word I just made up. Who knows? [05:06] >> Yeah. No, no. So, green ops is it's a [05:09] thing. It's a it's a developing thing. [05:10] It's far less mature than the world of [05:12] Philops. And the one thing about it is [05:14] you often hear cloud people particularly [05:17] say that green ops is cloud [05:19] sustainability. It's not at all. Never [05:21] has been. Um what it actually is, it's [05:24] the principle that covers the entire [05:25] operations of it. It's not just cloud. [05:28] It's data centers, infrastructure, [05:29] software development. It's how you buy, [05:31] what you buy, why you buy. It's how you [05:33] utilize. It's your operational [05:35] standards, your process, your [05:36] governance, your budgeting, the [05:37] everything to do with looking at your [05:39] world of IT through through the lens of [05:42] sustainability and the environmental [05:44] impact of your services. And it's an [05:46] interesting thing because it's it's an [05:47] enabler to ease the path to many other [05:50] decisions. So, it aligns perfectly well [05:52] with the well with the world of PHOPS. [05:54] Um because FOPS effectively at its heart [05:57] is really about the you know aligning [05:59] business value of cloud spend. So it's [06:01] it's helping you understand the [06:02] efficiency of what you spend why you [06:03] spend it bringing intelligence to it and [06:05] green ops in the world of cloud is just [06:07] aligning that from that sustainability [06:08] impact helping people understand [06:10] consciously you know their decisions [06:12] have an impact on the wider world that [06:14] these you know the cloud is not some [06:16] truly sustainable wonderful thing [06:17] lurking out there. It's actually just [06:19] another data center as everyone always [06:21] says and many of the cloud data centers [06:23] these days are actually some of them [06:25] although actually quite old now um some [06:27] of them operate a lot of legacy [06:28] infrastructure many of the cloud [06:30] providers actually have the same problem [06:31] that legacy IT had for many years so [06:34] actually greenups fits with anything [06:36] you're doing in IT whether it's in the [06:38] cloud in the data center or on your desk [06:42] >> is it a branding problem first of all I [06:44] think the cloud was probably a nefarious [06:47] floaty whimsical name that no one should [06:49] have come up with because it really does [06:52] physically take our mind away from this [06:55] idea of something that has weight and [06:58] permanence though nothing's more [07:00] permanent than a data cent's [07:02] environmental impact right [07:03] >> yeah sort of yes and but it's something [07:06] where measuring the impact has always [07:08] been very hard um and mainly because [07:11] actually the cloud providers have not [07:12] made it easy to measure the impact um [07:14] because their branding and their [07:15] marketing for many many years has been [07:18] predicated on the principle that we are [07:20] the most practically perfect, most [07:22] efficient sustainable oper operators in [07:24] the planet. So go with us because if you [07:26] do I mean you read their marketing all [07:28] day long it says things like we are 99% [07:30] more sustainable. It's nonsense. It's [07:33] completely [07:33] >> what they don't usually say then [07:35] anything right? [07:37] >> Yeah. So what what you find they [07:38] normally do is they uh they pick the [07:40] most practically perfect cloud [07:41] deployment that's never existed and they [07:43] balance that against the ugliest most [07:45] horrible infrastructure in a data center [07:47] and say look we've reduced it by 99%. [07:51] And the truth is no um because the truth [07:53] is always somewhere in between. You you [07:56] could have a perfect data center and a [07:57] clumsy cloud. You could equally have a [07:59] wonderfully wellconfigured cloud service [08:01] and an ugly data center. So it's [08:02] actually it's that whole sustainability [08:05] that you bring into it. So you got the [08:06] accountability of how you build, deploy, [08:08] and understand how you're delivering [08:09] your services. So no one's passing blame [08:11] on technology here. It's actually how [08:13] you consume the technology that's the [08:15] problem. [08:16] >> Fair enough. [08:18] Okay. So when we're talking about PHOPS [08:21] and greenups, we're talking about them [08:22] because one is kind of a good [08:26] parody for the other. So if you were [08:30] good at PHOps, you would [08:33] the the highest direct correlation that [08:36] you would be better for the earth, [08:37] right? Not that cloud is better for the [08:39] earth, but [08:40] >> I'd say [08:42] >> good PHOPS would be good for the earth. [08:44] >> Yes. No. maybe [08:46] um so and and the reason for all that is [08:48] I'm sure like Patrick would very much [08:50] know is there's many actions you could [08:52] take in the world of phops which [08:54] actually don't encourage efficient use [08:56] and consumption because things like um [08:58] spendbased commitments the cloud [08:59] providers are very good at convincing [09:01] you to make long-term commitments which [09:04] is just a very expensive way of buying [09:05] it they give you great discounts but [09:06] it's still very expensive and that [09:08] actually from a green perspective does [09:10] not encourage um efficient consumption [09:12] because there's no motivation to turn [09:14] something off below a threshold that [09:15] you've already committed to. And there's [09:17] areas where you you could have something [09:18] like a a more sustainable region. The [09:22] more sustainable regions tend to be the [09:23] newer ones and the newer ones tend to [09:25] come with higher unit costs. Not all the [09:28] time, but a lot of the time. So what you [09:30] have is a problem there is there are [09:32] directional things which are brilliant [09:33] in Phops with green ops. So things like [09:35] if you are um optimizing and right [09:38] sizing instances, awesome. You're [09:40] reducing your consumption, you're [09:41] reducing your power, you're reducing [09:42] your embodied emissions. You're turning [09:44] things off. So you're switching dev [09:45] environments off overnight. Brilliant. [09:47] Perfect for greenups. So So the truth is [09:50] some are, some aren't. You have to take [09:52] it in context. And and that's the [09:54] important thing. It's that lens of what [09:55] is the impact of my action. And and the [09:58] big thing is people always try and do [10:00] these things in isolation. So they have [10:02] the sustainability team or they have the [10:03] phops team and the problem is you end up [10:05] accidentally pulling in different [10:06] directions. It's actually that the where [10:09] greens works well is where it's not a [10:11] separate function. It's just a business [10:13] as usual thing. So it's something that [10:15] you build into your day-to-day work and [10:16] it becomes part of everyone's job not [10:18] just someone's job. [10:21] >> Okay. So then taking this to the next [10:24] step is cloudnative best practices, the [10:29] engineering best practices [10:31] better for the earth [10:34] or more secure [10:37] or more reliable. [10:39] >> Well, I'd let so I'd say from Patrick, [10:41] but I'd let him answer that one actually [10:43] because it's a if you are taking the [10:45] most efficient approach to building your [10:46] service and the most cost-ffective [10:48] approach at the start, then likely yes. [10:50] Patrick, that's definitely one for you. [10:53] >> I I think you know the the answer is it [10:56] depends. Like [10:58] >> most of the qu big questions in life, uh [11:01] it depends. And I' I've seen an [11:03] interesting trend lately of big [11:06] companies [11:07] deciding to move out of the cloud and [11:10] back to data centers. um primarily for [11:12] security and you know just understanding [11:16] that maybe it's a bit easier to manage [11:18] the fixed cost of a data center than the [11:20] variable cost of cloud. Um but I think [11:23] it it definitely does depend and even [11:27] down to the workload and what the [11:29] intention of of the application is that [11:31] you're trying to build. Um as well as [11:34] you know the [11:36] just knowing that you know Mark [11:38] mentioned things like scheduling [11:40] resources to turn off and on you know [11:42] that's something that's that's very [11:45] beneficial to the bottom line as well as [11:48] to the environment. Um, and for us at [11:53] Harness when they were developing [11:55] a capability called autostopping, they [11:57] they're treating it like [12:00] um, you know, modern conference rooms [12:02] where you walk into a room and the light [12:03] turns off on and you walk out, five [12:06] minutes later turns off, right? you're [12:08] you're enabling the functionality [12:11] without the user having to think about [12:12] it, without them having to manage the [12:15] the time schedule that something should [12:17] be off and on and trying to predict when [12:19] when it's going to be used and for that [12:22] you know that often is a a deterrent for [12:24] people using capabilities uh and trying [12:28] to be green. So I I think to me you know [12:31] the question really revolves around what [12:34] is the right approach for the company [12:37] for the workload and you know for our [12:39] end users and our bottom line. So it can [12:42] depend [12:43] >> and I think sorry I ask there is also an [12:47] elephant in the room when it comes to [12:48] things like cost as well. So one of the [12:51] biggest principles of of greenups is it [12:53] lets you get to leadership and because [12:55] often [12:57] it this is not true of everyone but in a [12:59] great number of organizations there's a [13:01] significant amount of waste. So [13:02] typically we find when when you strip [13:04] out all the waste that exists not the [13:06] obvious waste but all the waste it's [13:08] normally about 50 cents in every dollar [13:10] is wasted on it. And that isn't just the [13:13] obvious things you know the right sizing [13:14] and turning things off. actually things [13:16] like the the overdesign in the first [13:18] place, the the wasted resources during [13:20] development, the standardization that [13:21] hasn't been updated for many years. [13:22] There's lots of hidden waste that people [13:24] can't see. Um, and often we find that [13:26] actually leadership aren't actually [13:29] motivated that much by cost savings. [13:31] They all pretend they are, but actually [13:33] the reality is they're not actually that [13:35] motivated by it. So what you find is [13:37] going to them with a cost-saving message [13:39] on its own often doesn't get the result [13:41] you want because there's so many other [13:42] dependencies whether it's risk, [13:44] governance, compliance, time to invest, [13:46] the time required to do it, the [13:47] resources available and we find if you [13:49] take the sort of the environmental lens [13:51] to this um you don't catch everyone but [13:54] you catch enough of them and and you can [13:56] often win the hearts of minds of senior [13:58] leaders because the subject you're [13:59] talking about is something that vaguely [14:01] interests them. So it gets you into the [14:03] more strategic conversations because one [14:05] of the biggest successes we have in the [14:06] world of greenots particularly when it [14:08] comes to cloud is the whole and I I hate [14:10] the phrase but I use it anyway shift [14:12] left and what we mean by that is is [14:15] shifting from the world because [14:16] sustainability in greens often get stuck [14:18] in the world of reporting [14:20] >> and reporting changes absolutely [14:22] nothing. What you hear is people saying [14:23] that our our emissions are x tons and [14:26] great is that big is that small we don't [14:29] know. No one knows because the data is [14:30] so so poor and low quality. Um but when [14:33] you start shifting left then people go [14:35] into like optimization. And the problem [14:37] with optimization is optimization is [14:39] like um getting a getting a broom and [14:42] trying to brush away the tide sweeping [14:43] in over o over o over o over o over o [14:44] over o over o over o over o over o over [14:44] o over over your feet on the beach. You [14:46] know you might have a little bit of [14:47] success but tomorrow morning that tide's [14:49] coming back in. [14:50] >> Um but when you shift into avoidance [14:53] what you're doing is you're actually [14:54] trying to address the problem before [14:55] it's even happened. And the hardest [14:57] thing in the world of PHOPS is avoidance [14:59] is something that's really hard to [15:00] measure because you can't really measure [15:02] what you didn't do, but you can feel the [15:04] benefit later. And execs really struggle [15:06] with that understanding that you go, [15:07] well, so you're saying you've updated [15:09] our standards, our governance, our best [15:11] practice. And you, yes, and that's why [15:13] our bill has gone down. That's why our [15:14] environmental impact has gone down [15:16] because we made conscious decisions not [15:19] to deploy. So we did things like we you [15:22] know as standard we automated the [15:23] scheduling of things to be switched off [15:25] at night in dev environments. We we [15:26] standardized on the right sizing. We [15:28] chose different instances that had a [15:29] lower power consumption. We chose things [15:31] in the right region. So you you're [15:34] making that decision really early. So [15:36] and and often those decisions made early [15:39] they also had one unique thing in mind. [15:41] So Patrick, if I was to come to you and [15:42] say, "You've got a bunch of live [15:44] applications and I want you to switch [15:45] instance types from an Intel to a [15:47] Graviton processor type and I need it [15:49] done straight away." You will fight me [15:51] tooth and nail to not do that because of [15:53] the complexity, the risk, and the cost [15:54] involved in doing it. And you'll look at [15:55] me with cold dead eyes and you'll say, [15:57] "Here are the 9 million reasons why [15:59] you're dumb. You don't understand why we [16:01] can't do this." But conversely, if eight [16:05] months before when this was in the [16:06] ideation phase, I came to you and said, [16:08] "Look, we could do something [16:09] differently." Those decisions aren't [16:10] baked in yet. And when those decisions [16:12] aren't baked in, people are much happier [16:14] to change. And that change could be as [16:16] simple as, you know, we're going to [16:18] deploy this in Singapore, for example, [16:20] massively high carbon region, very high [16:22] impact. Why don't we put part of this [16:24] platform in Sweden during the [16:25] development phase because that's low [16:27] carbon energy. [16:28] >> You could do that in the early phase [16:30] because people are more willing to have [16:30] the conversation. Once it's live, you [16:33] try moving anything and you get you will [16:35] everyone will find a reason to say no [16:38] because you you've got like 97 business [16:40] units have to sign off the move. Five [16:42] different VPs have to allocate resource [16:44] to remapping applications. There's so [16:46] many other things that you that drag [16:47] you. So the whole principle of greenops [16:50] is that principle of use data you have [16:52] and shift it left to so adjust your your [16:55] processes, your standards. And the [16:57] biggest thing that we want people to do [16:59] and this is the hardest thing and we we [17:01] kind of talked about this previously [17:02] actually which is the the principle of [17:04] actually how you budget and how you [17:06] spend are one of the biggest drags and [17:08] cause of waste in enterprise IT today. [17:12] >> Explain that. [17:13] >> Um so so if if you take say um take an [17:16] average large enterprise they will have [17:17] a run budget and have a change budget [17:19] and and often they can't get a ch a run [17:22] budget for for subsequent years. So what [17:23] they do is the the the applicational [17:25] business unit bloat their change budget [17:27] as much as they can to grab as many [17:28] resources up front as they possibly can. [17:30] They encourage massive oversizing of [17:32] resources, massive overcommitment [17:34] because they don't know what they might [17:36] need in subsequent years and they cannot [17:37] guarantee they're going to have a run [17:38] budget. So because they can't guarantee [17:40] they're going to have a run budget, they [17:41] spend as much as they can in the crazy [17:43] time to go, well spend it whilst we can. [17:46] And you get that whole thing of we might [17:47] not have the budget next year, so let's [17:49] spend it now, which is just complete [17:51] nonsense. No wonder it never happens. [17:52] But people have the fear that it will. [17:54] And so because of that, you can't move [17:56] budgets between allocations. You can't [17:58] change pots. You can't change things [18:00] because you're not in charge of that [18:01] budget. So what happens at the back of [18:03] that is you end up actually with massive [18:04] overspend, which causes massive bloat [18:06] and massive waste. But if you could show [18:09] the leadership the impact of those [18:11] decisions, you'll find they're much more [18:12] willing to actually make changes to [18:14] reflect actually should we change how we [18:16] budget? Should we change how we allocate [18:18] funding? Because that also has an impact [18:19] on them higher up in the chain. [18:22] So how we've we've talked about a lot of [18:25] part of the problem is there's so many [18:26] stakeholders and you need to convince [18:29] really high up. So you need all those [18:31] pieces together to convince high up and [18:33] plan ahead. How are you I guess we [18:36] should take a second to like how are you [18:39] bringing all of this information [18:41] together? How are you creating a [18:42] centralized [18:44] view of everything to then have a for to [18:48] have the capability to talk across these [18:51] different departments? Is it a platform? [18:54] What is the answer to then [18:57] be able to then strategically make this [18:59] argument moving forward? [19:01] >> So [19:03] I have a fairly blunt perspective on [19:05] this [19:05] >> which is [19:07] >> the answer is never another tool. Um, it [19:10] it has it has a problem to say let let's [19:13] throw a dashboard at this problem. Um, [19:15] and and they'll build things in [19:17] isolation. So from my perspective, we [19:18] always see particularly large [19:20] enterprises, they go, we're going to [19:21] build our climate dashboard. We're going [19:22] to build our central sustainability [19:24] dashboard. And they get so excited over [19:26] doing it. They build this big central [19:28] thing. They consolidate all the data and [19:30] they bring it in there. So they'll do [19:31] like cloud data enrichment. They'll [19:33] start getting a baseline of the cloud [19:34] services and they'll be really proud. [19:36] And they build this separate thing. And [19:38] the problem is within about 3 months [19:40] everyone's got bored of it and no longer [19:41] goes to it. So actually what you need to [19:44] focus on is improving the quality of the [19:45] data you have and then get that data [19:48] into the tools and systems that you're [19:49] already working in. So go into every [19:51] function that every area whether it's [19:53] particularly like phops is a brilliant [19:55] example. So you get this data into your [19:57] phops view get it into those tools that [19:59] the people are using already. So take it [20:01] to them and get that same data into the [20:03] observability tools get it into the [20:04] devops tools. You have to and because [20:06] there is no one KPI to rule them all. [20:09] People like to try and come up with one [20:11] but there just isn't one and that's [20:12] because everyone looks at this problem [20:14] through a different lens. But what [20:16] >> I think to that point Sorry Mark [20:19] >> Yeah. I think to that point it's you're [20:22] you're absolutely right. It's it's not [20:24] you know one KPI to rule them all. [20:26] There's not one solution for this. [20:29] You have to force the hard work for the [20:31] people who can make the decisions to get [20:33] in the room and say, "Okay, what is the [20:35] problem we're trying to solve before we [20:38] decide let's get another tool or let's [20:40] spin up a dashboard or let's, you know, [20:42] whatever the solution is, right? They [20:45] have to agree on what is the problem [20:48] that they are going to commit themselves [20:50] to solve and to do the hard work and [20:52] when things get really hard to push [20:54] through and and and take it to the end." [20:58] Um, and for different companies that [21:00] might be completely different [21:01] priorities. Some might be interested in [21:04] reducing cost because their stock is [21:06] underwater or they might, you know, be [21:09] interested in in environmental impact. [21:12] And so, you know, they don't care so [21:15] much about cost. I mean, it would be uh, [21:17] you know, generally you would say they [21:18] correlate where if you're driving down [21:20] your environmental impact, your cost [21:22] would come down um, in most cases, but [21:24] not all. Um I I think that's really you [21:27] have to start by examining what is the [21:30] problem we're here to solve before you [21:32] decide on on what tool or what which way [21:34] forward. [21:35] >> Yeah, the the unit economics of it are [21:37] very important because you need to find [21:39] you need to find the metric that works [21:40] for the persona that you're talking to [21:42] >> and you need to visualize that in a way [21:44] that works with how they currently think [21:45] and behave today because bringing in [21:47] endless new KPIs, OKRs of metrics never [21:50] really gets you very far because if [21:51] they're not stuck in them already, it [21:53] just won't work. they won't use them. [21:55] And you also have to show how it's [21:56] aligned with what they actually care [21:58] about. Often you'll find there'll be um [22:00] you know the major strategic initiatives [22:02] for the year. There'll be like a data [22:03] center modernization, a cloud [22:05] transformation, an application [22:06] modernization, all the lovely phrases we [22:08] have. If you can show how you're going [22:10] to align this with them to help that [22:11] that be a success, you've got a much [22:13] greater chance of them jumping on board. [22:15] Because the one thing we have found is [22:18] um when you use environmental metrics, [22:20] it's a really good motivator for enough [22:22] people to want to get involved and [22:24] understand more because what you're [22:25] doing is you're giving them a different [22:26] lens to view a problem that they already [22:28] know exists because everyone in it knows [22:30] where the waste is. Often the problem is [22:32] not finding the waste. The problem is [22:34] getting people to get out the way of you [22:36] eliminating the waste in the first place [22:38] because it isn't as simple as everyone [22:39] thinks. There are so many other [22:40] dependencies and roles and people [22:42] involved in doing it. But if you show [22:43] them their impact. So for example, we [22:45] often see things like um application [22:47] demise projects that are massively [22:48] stalled. Whereas services that are just [22:50] left running, often they're just left [22:53] running because the department whose [22:55] budget it was, it's already committed [22:56] for them. They don't get any money back [22:58] if they switch it off. There is no [22:59] saving for them. So they don't they [23:00] really don't care about it. But what [23:02] what happens is they have a cost [23:04] associated to them of terminating [23:05] something. But they actually have to [23:07] allocate resource on their side to [23:08] switch things off and terminate stuff [23:10] and they don't get any money back. So [23:11] they can't really be bothered to do it. [23:13] But if you show them the environmental [23:14] drag associated with, you know, every [23:16] month that you're not switching this [23:17] off, it's another five tons of carbon, [23:19] for example, they're more likely to be [23:21] motivated to get involved. And also, we [23:23] have to be conscious that carbon doesn't [23:25] motivate everyone. Often you have to [23:26] find another. [23:27] >> Yeah, I was going to I was going to push [23:29] back on that just to play devil's [23:30] advocate. Not to say that [23:33] >> yeah I car I mean listen I am [23:37] >> I would put myself in the camp of [23:38] environmentally conscious but just to [23:40] push back and and and hear more of your [23:42] perspective Mark because you're closer [23:44] to it than I am. Carbon feels like a [23:46] more abstract concept to me than a [23:49] dollar. And so I don't think that [23:53] engineers or even leaders would care as [23:55] much about carbon impact and that would [23:57] motivate them to to make the change. How [23:59] do you see that playing out or or like [24:03] maybe it's not the only I'm sure it's [24:06] not a binary like what else? [24:09] >> Yeah. So it's actually [24:10] >> more about [24:11] >> Yeah. So so that's where when you get [24:12] into it there's a lot of heavy lifting [24:13] you have to do to get people to the [24:15] understanding of how big is big because [24:17] when you just give the metric like like [24:19] a ton of carbon that's a really abstract [24:20] thing. So you have to quantify in terms [24:22] that they understand and people always [24:24] come up with terms like flights from [24:26] Philadelphia to New York or flights from [24:28] Barcelona to Glasgow that kind of thing. [24:30] You get this kind of thing. So you have [24:31] to find a metric that lands and often [24:33] also carbon is one sometimes that you [24:35] have to separate out because it's so [24:36] abstract and you have to find other [24:38] things as well. So often energy is used [24:40] now as well even though energy is only a [24:42] subset of emissions a very small subset [24:44] but they still want to quantify that to [24:45] have you know what is our carbon [24:47] efficiency per dollar spent for example [24:49] is a KPI we often see so they can start [24:52] basically what levers can they pull and [24:53] that's actually part of bringing it into [24:55] the world of phops as well is aligning [24:57] it with existing unit economics that are [24:59] being utilized and just show a different [25:01] lens on the number [25:03] >> then there's also the the really [25:05] interesting thing we're finding because [25:06] you mentioned pe about people starting [25:08] to move from the cloud back to their own [25:10] data centers. We're starting to see that [25:11] as a real trend, particularly for [25:13] services like AI. And the reason is is [25:16] because AI is so ridiculously expensive [25:18] in the cloud that actually people can't [25:20] afford to run production systems and [25:21] they want the control because in the [25:23] cloud um if you you know if you oversize [25:26] something on an AI platform, it's not a [25:27] million dollar mistake, it's a $30 [25:29] million mistake. [25:30] >> So So actually you get greater controls [25:33] and limitations and you go back to the [25:34] old world of everyone can see it when [25:36] you're running out of capacity. It's [25:37] really under your control and your ey [25:39] line. [25:40] >> Yeah. Um on the staying on the topic of [25:43] carbon as the sort of currency in this [25:47] equation when I was I was doing a little [25:48] bit of reading and research in prep for [25:52] for our talk today and um you know we've [25:56] we've at harness have studied the cost [25:59] of AI like the actual dollar cost of AI [26:01] because we have customers and prospects [26:03] who ask us what does our pinops platform [26:05] do for the cost of AI and it does a lot [26:07] and I won't go into all those details [26:09] the the point of [26:11] of mentioning that is [26:14] to to the typical company out there that [26:17] spends hundreds of millions of dollars [26:19] on cloud infrastructure every year they [26:21] spend less than 1% less than I think [26:23] even half a percent on AI [26:25] >> but it's just this the mind share of AI [26:28] is so great these days obviously and [26:32] like when I look at my you know when I [26:34] book a a flight to go somewhere now my [26:36] ticket comes and it tells me how much uh [26:39] you know how much carbon when I'm [26:40] contributing to the environment for this [26:43] flight. Again, it's such an abstract [26:45] concept to me, but when I was doing the [26:47] research for for this, you know, I I [26:51] wanted to know how wasteful is AI. And [26:53] the one interesting thing that I found [26:56] was that um and I'm just checking some [26:59] of my notes. AI data centers are are [27:03] expected to consume more six times more [27:08] fresh water than the entire country of [27:10] Denmark. [27:11] >> That all of a sudden to me is like [27:13] that's a lot of water. Like if I how [27:16] many bottles of water I would have to [27:18] consume that to matter like [27:20] >> but motivate you to not use AI in any [27:23] way. Are you rethinking like oh maybe I [27:26] shouldn't use AI right now? [27:28] >> Maybe. Yeah. I mean, in the moments when [27:31] you can avoid it, I mean, you do a [27:32] Google search and it's gonna it's going [27:33] to run the, you know, the AI summary at [27:35] the top regardless if you want it or [27:37] not. But you're right, like that's the [27:39] thing. So I think maybe the unlock there [27:41] is is and Mark you're you're sort of [27:43] alluding to it is get it to a metric or [27:47] just a a data point that people [27:50] understand [27:52] >> because if we just talk carbon I just [27:55] feel like it's going to go over most [27:57] people's heads. [27:59] >> Can I say other things because I I'm [28:03] going to try to bring the voice of Ann [28:04] or imbue my Anna. She's still trying to [28:06] get online, which I can't imagine how [28:08] frustrating that is. So, thank you for [28:10] your patience and everyone that to see [28:14] her. So, we were talking before um this [28:17] call about how to make leadership care [28:20] and [28:22] it's not just about money, cloud spend, [28:27] or the environment. There's other ways [28:29] that maybe might pull that level lever [28:32] more that will end up helping all those [28:35] things. [28:36] But we talked about before security or [28:38] resiliency because right now if you have [28:42] like what is it 50% of services might be [28:44] zombies or we don't even know how many [28:46] zombie services [28:48] >> like these are huge risks if you're not [28:50] maintaining things if you're not [28:51] upgrading these are other ways. So [28:54] >> absolutely yes there's a compliance risk [28:56] there's a governance risk there's all [28:57] sorts of risks like that that come into [28:58] it that come diving right in and that [29:02] also is a you also have concentration [29:03] risk factors as well. So we're [29:05] increasingly starting to see [29:06] particularly very large consumers are [29:08] looking at the risk of of how many of [29:10] their competitors are also running in [29:12] the same cloud regions as them. Do they [29:13] want to be in that same cloud region? [29:15] Should they be diversifying services [29:17] into different regions? Um and then they [29:19] have to look at availability of [29:20] resources because that's particularly [29:21] relevant for in the world of AI where [29:24] actually you want to be in a certain [29:25] region but you just can't literally [29:26] cannot get the resource um because they [29:28] they can't build it fast enough at the [29:30] moment because the rate at which people [29:31] are trying to consume it. So actually [29:32] you have to make very conscious [29:33] decisions and there are also then issues [29:36] from a corporate perspective where you [29:38] look at things like what are there risks [29:40] to water usage in the region which [29:41] you're operating. Are you actually [29:43] adding to the problem? So you're not [29:44] being a good corporate citizen by using [29:46] a data center in region X. It doesn't [29:48] mean the data center is to blame. It [29:49] just means you should be reducing the [29:50] workload and putting it somewhere else. [29:55] >> Yeah. what Ann had said when we talked [29:57] about because I was just it's very [29:59] frustrating to see continual layoffs [30:02] when you could cut 50% of your cloud [30:04] cost and then if you have to do layoffs [30:08] take more time to do them more sensibly. [30:10] So my understanding is that leadership [30:14] has to care more than and environment [30:16] and even spending may not do that. But [30:20] if you invest in operations, Ann was [30:24] trying to explain to me we're talking [30:25] about this because I was just very [30:27] frustrated and and said like investment [30:30] means building the skills of your [30:32] operations teams in resilience and [30:35] security because that will have a [30:37] knock-on effect of cutting waste. So [30:39] it's also about building into your staff [30:42] these resiliency, security things that [30:46] you have to compliance. [30:48] >> Yeah. And so [30:50] one of the hardest things about the work [30:51] that we often have to do is when we talk [30:54] about that shift lefting of working with [30:56] leadership is we effectively have to [30:58] tell them that their people are not the [31:00] problem, their technology is not the [31:02] problem, they are the problem. how to [31:05] articulate that phrase to leadership [31:07] which is um big decisions can't be made [31:10] because they're in the way of those big [31:11] decisions. So they're the ones who [31:12] aren't giving the time, the resource, [31:14] and the budget to do it. And they're the [31:15] ones, if you take that example you gave [31:16] of layoffs where despite a huge amount [31:19] of waste on spend, people are saying, [31:21] "We're still going to have to cut [31:21] heads." And you could say, "But hang on, [31:24] I could save half a billion dollars over [31:26] here a year. There's no need for you to [31:28] cut heads." But simply because someone [31:30] at a CFO level is unwilling to change [31:31] how they budget or allocate funds means [31:34] actually you end up having to cut a huge [31:35] amount of headcount simply because one [31:37] exec somewhere does not want to change [31:39] because they don't understand the impact [31:40] of their decision. So there is a certain [31:42] level of problem here where we have a [31:44] lack of accountability at an exec level [31:46] where they're not they don't effectively [31:48] have their eye on the detail because I I [31:50] often go to very large organizations and [31:52] you find they don't have a clue for [31:53] example um what their cloud spend is [31:56] where it is they don't understand the [31:57] level of waste they've got. They don't [31:58] know how, you know, the the data center [32:00] investment. They don't know the server [32:02] utilization. They don't know these core [32:03] metrics that they really should know [32:04] because this isn't just a cloud problem [32:06] by the way because we're sort of focused [32:07] on that because obviously that's put on [32:08] the subject. But in data centers, [32:10] there's also epic amounts of waste that [32:12] happens as well. So it's not it's not [32:13] like data centers are perfect. You get a [32:15] lot of really low utilization in on [32:17] premise data centers in the same way [32:18] where things get built, deployed, run, [32:21] not turned off. The exactly the same [32:22] problems exist. So it all comes down to [32:25] giving people the time they need to make [32:27] the decisions they need because [32:28] engineers don't deliberately make [32:30] inefficient decisions. They want to [32:32] develop things efficiently and consume [32:34] them efficiently, but actually the [32:35] complexity of the organization gets in [32:37] the way of those decisions often because [32:39] they're stuck down to what they can [32:40] change. [32:41] >> Yeah. I think the challenge with you [32:43] know with the the evolution of cloud [32:46] infrastructure and the big migration to [32:49] cloud was that you know and and sort of [32:52] in my opinion one of the leading causes [32:54] of inefficiency in the cloud is just [32:56] that the majority of the engineers who [32:59] were moving things to the cloud were [33:01] steeped in data center concepts and [33:05] infrastructure and like fixed capacity [33:08] and basically you stand up a server and [33:11] you get the thing and you don't get the [33:12] question and there you know there was a [33:14] time before there were virtual machines [33:16] and the ability to load balance [33:18] capacity. Um, [33:21] that's that's just the evolution of it. [33:23] And I think [33:26] we need a [33:28] we need to be able to say in front of [33:31] our peers, I don't know the best way to [33:35] do something, [33:36] >> but I'm willing to learn [33:39] as long as, you know, I don't get thrown [33:41] under the bus or discarded because I [33:44] don't know. today like we're seeing [33:47] better efficiency in the cloud than we [33:49] saw five years ago. We're seeing [33:52] um [33:53] you know we're but we're seeing a shift [33:55] back to data centers as well for other [33:57] reasons. So I think it's just a matter [33:58] and I think probably it's the same in [34:01] the greenoff space is there's a lot we [34:03] don't know collectively and individually [34:06] and that that really I feel like if we [34:09] could take down that sort of stigma of [34:12] not knowing then we might be able to [34:16] move forward a little better a little [34:18] more. You know, I would absolutely agree [34:20] because there is in um particularly in [34:21] the world of the kind of the [34:23] environmental data that people are [34:24] using. None of it is none of it is [34:25] perfect. All different levels of quality [34:28] and you often spend too much time [34:29] arguing about the quality of the data [34:31] than you do about fixing the problem. [34:32] It's just a case of how how big is a [34:34] unit of something. We know there's [34:36] waste. Let's get that unit which which [34:38] then sort of becomes a problem for [34:40] people to then try and actually fix and [34:41] address. And it's how you how you then [34:44] take that data and manipulate it and get [34:46] it into a way that people in all the [34:47] different functions who are involved. [34:48] And that goes anything from kind of [34:50] supply chain through engineering through [34:52] to leadership. There's there's no one [34:53] single place here. Everything everything [34:55] has to answer the question. [34:58] So I see that I see there's there's [34:59] there's a question in chapter is an [35:01] interesting one. So caring about carbon [35:03] is a cultural thing. Europeans care and [35:05] Americans don't. So [35:06] >> I mean I would never lump Europe is a [35:10] continent with as many people as the [35:13] United States and both have diverse [35:15] opinions. So I do a I do want to take as [35:18] an American living in Europe uh that [35:21] stance. [35:23] >> Yes, [35:23] >> I that does that is an individual thing. [35:26] But I guess what we should take this as [35:29] how [35:31] because I don't know that we can say [35:33] that the top 100 companies in Europe are [35:37] doing the due diligence but certainly [35:39] there is more regulation specifically in [35:41] the EU to do that and measure it. I [35:45] think we could say I wouldn't say it's a [35:47] cultural thing that or a binary in that [35:49] way. I would just jump in to say [35:51] >> yeah. No, I would I would I would say I [35:53] I would tend to agree because I I work a [35:55] lot with um American firms and what we [35:58] find is the you are right there's more [36:00] of a balance of of organizations in the [36:03] US who don't care um because their [36:06] priority is actually about growth, [36:07] stability and that that side of things [36:10] and there's more companies in Europe who [36:12] do care but it's not that black and [36:15] white. It comes down to actually I'd say [36:17] there's probably if I was to separate it [36:19] actually between the workforce and the [36:20] leadership very different thing then [36:23] leadership in the US definitely does not [36:26] care right now there are many reasons [36:28] for that but leadership does not care [36:30] >> they have no motivator it's different [36:32] the reason European leadership or [36:34] leadership in Europe or the EU may care [36:37] because they are like financially [36:39] motivated by 2027 to comply to measure [36:43] to report which is [36:46] completely different cultures [36:48] governmentally. [36:49] >> Yeah. But there is there is also an [36:52] interesting second question that's come [36:53] in there which is about the energy price [36:55] change impact. That is actually a a very [36:58] good motivator in the world of greenops. [37:00] Now that's why I mentioned other metrics [37:02] other than just carbon because if you [37:03] start building energy in people start [37:05] looking at the energy impact of their [37:06] consumption and comparing and [37:08] contrasting on premise with the cloud [37:10] estimations of what they might be [37:11] consuming and realizing actually there [37:13] is a huge impact because the unit [37:14] economics involved in energy pricing. [37:16] Energy is becoming a substantial [37:18] proportion of cloud unit costs and and [37:21] you're starting to see more more cost [37:22] spikes on a perunit basis just simply [37:24] because the cost is going through the [37:25] roof. And that's where you you're now [37:27] seeing cloud cloud purchasers buying um [37:30] you know energy commitments longer and [37:31] longer ahead and also that's one of the [37:33] reasons why some of them are investing [37:34] in renewable energy so they can get more [37:36] more more of their energy supply under [37:38] their control as well. So geopolitically [37:40] as well there's that challenge to [37:42] handle. So it isn't just the carbon [37:43] impact it's the energy sourcing and the [37:45] other big problem that all big [37:46] hyperscalers have which is they can't [37:48] build fast enough because they can't get [37:49] the connectivity to the grid. they [37:51] literally cannot get the power they need [37:53] in certain regions and they're running [37:55] out and so there's a societal impact of [37:56] that as well. So we're starting to see [37:58] the impact of data centers are causing [37:59] an impact on the local societies [38:01] surrounding them as well where they're [38:03] getting priority over energy supply over [38:05] local societies including things like [38:07] hospitals and other critical services. [38:10] So there is a market fluctuations is an [38:12] excellent driver to drive the motivator [38:14] to optimize to be as frugal as possible [38:16] because the whole principle of frugal by [38:18] design is something which we're now [38:20] starting to see because we saw a very [38:22] interesting test in a UK organization. [38:24] What they did was um because in their [38:26] old world where they were in a data [38:28] center they were limited by the [38:29] physicality of what they had because if [38:31] someone wanted more resources someone [38:33] had to sign off and buy a new server as [38:35] simple as that. They couldn't just go [38:36] and consume it. when they went into [38:38] world of cloud, they weren't very [38:39] mature. So it spiraled very quickly [38:41] because the control went to the edges [38:42] and everyone could do anything because [38:43] they were all cool kids. Um but what [38:46] they so so all their costs massively [38:48] spiked despite the cloud providers [38:50] telling them they were going to save [38:51] money by moving to the cloud. Harsh [38:53] reality they didn't their IT spend went [38:55] up by 80%. But what they did was they [38:57] did a test in their ecosystem which was [38:59] they went in and said right we're [39:00] assuming we cannot we cannot consume any [39:02] more resources as being a geopolitical [39:04] event. we cannot increase our cloud [39:05] spend at all because there are no more [39:06] resources available. [39:08] And suddenly the funniest thing was [39:10] everyone's mindset shifted from let's [39:12] switch on and test everything and do new [39:14] things to actually what can we optimize? [39:16] What can we switch off? What what are we [39:17] not using? What services are pointless? [39:20] What what what should we put on the [39:21] backlog? What should we not do? And you [39:23] know on the from a business perspective [39:26] nothing changed. [39:27] >> There was absolutely there was no [39:29] impact. So actually cloud was not [39:31] delivering the value they thought it [39:32] was. It was just a thing that delivered [39:34] the platforms they were running. [39:36] >> It's about that's it's about designing [39:38] for resiliency in and uncertainty. Uh [39:42] just something I want to mention and [39:44] will be joining in the Slack community [39:46] so we can do asynchronous conversation. [39:50] Frustrating that she couldn't join but [39:52] especially for her of course after we [39:54] did all this prep. But I would [39:56] definitely follow An Curry on LinkedIn [39:58] because she's been playing around with [39:59] this idea of solar native. Uh which does [40:02] not mean just solar power, but the idea [40:05] that maybe you can do workloads that [40:07] automatically shift to when whatever [40:10] renewable energy is available is [40:12] available. Uh I just want to briefly as [40:14] we wind down give uh our one of our last [40:18] questions that I think is we could do an [40:21] entire webinar on this but [40:24] >> the reader the viewer wrote it was [40:28] mentioned that a new dashboard isn't the [40:30] answer neither is optimization like [40:32] right sizing your instances. What is the [40:34] p panel's first suggested step for any [40:38] organization trying their hand at [40:40] something like green ops for the first [40:42] time? [40:43] >> B [40:44] >> don't [40:45] >> a new dashboard is not the answer. [40:47] >> Yeah. What is the first [40:49] >> optimization and right sizing definitely [40:51] is so doing the things you know you [40:52] should already be doing but using [40:54] environmental data to assist you with [40:56] that. So don't do it in isolation. So [40:58] don't just have a separate team doing [40:59] it. work with the Phops teams, work with [41:02] the engineering teams, bring the [41:03] visibility of the data into that stream. [41:05] Don't do it in that isolation. And if [41:07] you can build it into the dashboards you [41:09] already have so what what I was saying [41:10] there was don't go and buy another tool [41:12] just for the sake of it that people [41:14] won't use. Build it into your existing [41:16] systems of record and services that [41:17] people are comfortable with because [41:19] engineers hate being told here's another [41:21] thing you've got to go and log into to [41:22] do something. They just want to live in [41:24] the tools they're comfortable with. If [41:26] you do that, it works so well because [41:28] that it's in their ey line. they can see [41:30] it alongside the metrics they're using [41:31] as well. [41:33] >> Okay, Patrick, last word. What would you [41:35] do as the first step? [41:37] >> I agree. I think it's meet your [41:40] engineers where they are uh with the [41:42] data that matters and and encourage them [41:45] to continue on their right sizing [41:48] journey. And you've got to sort of, [41:51] you know, humans are motivated by [41:53] different factors. And some might be [41:55] motivated by seeing the bill go down. [41:58] Others might be motivated by carbon [42:00] impact. Others might be motivated by [42:01] something completely different. Um so [42:03] it's really talking to your constituents [42:07] and understanding what it is that [42:08] motivates them and you know what would [42:11] be helpful to them or not. Uh I think [42:13] Mark hit on most of it which is you know [42:15] I think FinnOps has a lot to contribute [42:17] to green ops. So I would say it's [42:20] continue doing what you're doing in that [42:22] space, but just bring the awareness to [42:25] we're we're impacting much more than [42:27] just our cloud build. There's [42:29] environmental impacts, there's societal [42:31] impacts, etc. You know, Mark talked [42:33] about data centers taking priority over [42:35] local citizens for electricity. Uh you [42:39] you could show impact there as well. [42:43] >> Absolutely. And just to give a little [42:45] tease to join us in the Slack community, [42:48] I've already added it there. Uh Ann's in [42:50] there. So we we continue in asynchronous [42:53] lead dev is a wonderful community where [42:55] you can continue that conversation. Uh [42:58] I've shared in the Slack and those that [43:00] join us. But uh just to tease there was [43:03] the question of who are the right people [43:05] to tackle cloud resource usage and how [43:08] can they find the right levels levers to [43:11] start tackling it and [43:14] >> if you're not sure we need transparency [43:17] in our organizations [43:19] >> but I think there's a lot of work to do [43:22] it is urgent [43:24] >> as is and it's tied in with the urgency [43:27] to make your systems more resilient. you [43:31] cross purposes between compliance, [43:34] security, business, technology all need [43:37] centralized viewing, centralized [43:39] communication, talking together. So, I [43:43] wish everyone luck. I really want to [43:45] thank Mark and Patrick and an who is [43:47] joining us already in the Slack. We will [43:49] continue the conversation. [43:52] This is an important one and I hope it [43:55] may have brought more questions to you [43:57] than answers, but they're questions to [43:59] bring back to your team and your [44:01] organization. So, thank you so much. [44:04] Thank you harness for supporting this [44:05] and lead dev. Thank you Mark and [44:06] Patrick. [44:07] >> And good luck y'all. We're just on the [44:09] start of a really urgent important [44:11] journey. Thank you. [44:12] >> Thank you everyone.  📄 **Full text:** ![[Full Document Contents/Articles/How Building Greener Software Cuts Your Bottom Line|How Building Greener Software Cuts Your Bottom Line, YouTube]] ## Highlights - more sustainable. It's nonsense. It's completely >> what they don't usually say then anything right? >> Yeah. So what what you find they normally do is they uh they pick the most practically perfect cloud deployment that's never existed and they balance that against the ugliest most horrible infrastructure in a data center and say look we've reduced it by 99%. And the truth is no um because the truth is always somewhere in between. You you could have a perfect data center and a clumsy cloud. You could equally have a wonderfully wellconfigured cloud service ([View Highlight](https://read.readwise.io/read/01kmjrnsf7vntfgztq4wpzpyvb)) - Tags: 👻 ai highlighted - and an ugly data center. So it's actually it's that whole sustainability that you bring into it. So you got the accountability of how you build, deploy, and understand how you're delivering your services. So no one's passing blame on technology here. It's actually how you consume the technology that's the problem. >> Fair enough. Okay. So when we're talking about PHOPS and greenups, we're talking about them because one is kind of a good parody for the other. So if you were ([View Highlight](https://read.readwise.io/read/01kmjrnsg6rj2cr0b5b5nmqqyt)) - Tags: 👻 ai highlighted - you know the major strategic initiatives for the year. There'll be like a data center modernization, a cloud transformation, an application modernization, all the lovely phrases we have. If you can show how you're going to align this with them to help that that be a success, you've got a much greater chance of them jumping on board. Because the one thing we have found is um when you use environmental metrics, it's a really good motivator for enough people to want to get involved and understand more because what you're doing is you're giving them a different lens to view a problem that they already know exists because everyone in it knows ([View Highlight](https://read.readwise.io/read/01kmjrnskd16fsc3vttjp6mn87)) - Tags: 👻 ai highlighted - lens on the number >> then there's also the the really interesting thing we're finding because you mentioned pe about people starting to move from the cloud back to their own data centers. We're starting to see that as a real trend, particularly for services like AI. And the reason is is because AI is so ridiculously expensive in the cloud that actually people can't afford to run production systems and they want the control because in the cloud um if you you know if you oversize something on an AI platform, it's not a million dollar mistake, it's a $30 million mistake. ([View Highlight](https://read.readwise.io/read/01kmjrnsr4acdepag8tagwevz0)) - Tags: 👻 ai highlighted - when you could cut 50% of your cloud cost and then if you have to do layoffs take more time to do them more sensibly. So my understanding is that leadership has to care more than and environment and even spending may not do that. But if you invest in operations, Ann was trying to explain to me we're talking about this because I was just very frustrated and and said like investment ([View Highlight](https://read.readwise.io/read/01kmjrnsy476mp409vb9pm86yz)) - Tags: 👻 ai highlighted - utilization. They don't know these core metrics that they really should know because this isn't just a cloud problem by the way because we're sort of focused on that because obviously that's put on the subject. But in data centers, there's also epic amounts of waste that happens as well. So it's not it's not like data centers are perfect. You get a lot of really low utilization in on premise data centers in the same way where things get built, deployed, run, not turned off. The exactly the same problems exist. So it all comes down to giving people the time they need to make the decisions they need because engineers don't deliberately make ([View Highlight](https://read.readwise.io/read/01kmjrnt4vxsw8k98jxyjccq4r)) - Tags: 👻 ai highlighted - our peers, I don't know the best way to do something, >> but I'm willing to learn as long as, you know, I don't get thrown under the bus or discarded because I don't know. today like we're seeing better efficiency in the cloud than we saw five years ago. We're seeing um you know we're but we're seeing a shift back to data centers as well for other reasons. So I think it's just a matter and I think probably it's the same in ([View Highlight](https://read.readwise.io/read/01kmjrntdv3rjw39wr320g7gmk)) - Tags: 👻 ai highlighted - something completely different. Um so it's really talking to your constituents and understanding what it is that motivates them and you know what would be helpful to them or not. Uh I think Mark hit on most of it which is you know I think FinnOps has a lot to contribute to green ops. So I would say it's continue doing what you're doing in that space, but just bring the awareness to we're we're impacting much more than just our cloud build. There's environmental impacts, there's societal ([View Highlight](https://read.readwise.io/read/01kmjrntt53crqs50aqa9r1drr)) - Tags: 👻 ai highlighted