--- title: P99 CONF Speaker Spotlight: Madelyn Olson | daily.dev description: A conference speaker spotlight interview with Madelyn Olson, Valkey co-creator and AWS Principal Software Engineer, previewing her P99 CONF talk on the... canonical: https://daily.dev/posts/p99-conf-speaker-spotlight-madelyn-olson-zh968yypl twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: P99 CONF Speaker Spotlight: Madelyn Olson | daily.dev og:description: A conference speaker spotlight interview with Madelyn Olson, Valkey co-creator and AWS Principal Software Engineer, previewing her P99 CONF talk on the... og:url: https://daily.dev/posts/p99-conf-speaker-spotlight-madelyn-olson-zh968yypl og:image: https://api.daily.dev/og/posts/zH968yYPL.png og:image:alt: P99 CONF Speaker Spotlight: Madelyn Olson og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # P99 CONF Speaker Spotlight: Madelyn Olson **[P99 Conf](https://daily.dev/sources/p99conf)** · 5 min read · 5 upvotes · 1 comments ## Summary A conference speaker spotlight interview with Madelyn Olson, Valkey co-creator and AWS Principal Software Engineer, previewing her P99 CONF talk on the vertical scaling wall and how Valkey addresses it through both scale-up and scale-out approaches. She discusses ongoing work on data tiering in Valkey to offload from DRAM to NVMe storage, driven by AI inference KV caching workloads and DRAM cost pressures. She also mentions recent internal rewrites — replacing a skiplist-based ordered index with a B+ tree and rebuilding the main hashtable as a Swiss table — and recommends reading design discussions in real open source codebases to learn about performance tradeoffs. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is data tiering in Valkey and why is it being built now? Data tiering in Valkey moves some data from DRAM to NVMe storage instead of keeping everything in memory, a shift made practical because NVMe latency has dropped and bandwidth has climbed while DRAM remains the biggest cost driver in Valkey deployments. Demand comes from AI inference KV caching workloads with extreme throughput and capacity needs, plus teams seeking cost savings. AWS previously built a similar feature for ElastiCache and is now building an open source version for Valkey. _Anyone tracking cost-driven storage tradeoffs can follow Valkey's tiering work as it develops on daily.dev._ ### Why does Valkey support both scaling up and scaling out instead of picking one? Valkey supports both because each approach has distinct failure modes: scaling up on one large node creates a bigger blast radius, longer replication times, and wasted spend when hardware doesn't match the workload, while scaling out has traditionally suffered from slow slot migration and disruptive rebalancing. Command execution remains single threaded in Valkey for now, a deliberate choice discussed alongside the metrics recommended for deciding when to scale. _Teams weighing scale-up versus scale-out tradeoffs can track how Valkey's approach evolves on daily.dev._ ### What internal data structure changes has Valkey made recently? Valkey replaced its original ordered index, a skiplist that had been part of the codebase since the beginning, with a B+ tree, and rebuilt its main hashtable as a Swiss table. Both changes took established designs and adapted them to Valkey's specific constraints of being memory hungry while avoiding latency stalls, since the project aims to prevent cache timeouts. _Developers studying database internals rewrites can follow projects like Valkey's design changes on daily.dev._ ## Community discussion Top comments from developers on daily.dev. **@trevorsuna** · 0 upvotes > NVMe tiering feels like a practical response to DRAM cost, especially for large AI inference caches. The important measurements will be tail latency during promotion and eviction, not just capacity gains. The B+ tree and Swiss table rewrites also sound like excellent real-world performance case studies. ## Similar posts on daily.dev - [\[Video Podcast\] Improving Valkey with Madelyn Olson](https://daily.dev/posts/video-podcast-improving-valkey-with-madelyn-olson-bjpwcn9zl) · InfoQ · 0 upvotes · 0 comments - [Valkey just had a 17x year. Its lead maintainer still doesn’t want Redis to die.](https://daily.dev/posts/valkey-just-had-a-17x-year-its-lead-maintainer-still-doesn-t-want-redis-to-die--hymzgle9p) · The New Stack · 2 upvotes · 0 comments - [Books by P99 CONF Speakers: AI to Sci Fi](https://daily.dev/posts/books-by-p99-conf-speakers-ai-to-sci-fi-b6rc8hdph) · P99 Conf · 0 upvotes · 0 comments --- Tags: [#performance](https://daily.dev/tags/performance), [#backend](https://daily.dev/tags/backend), [#distributed-systems](https://daily.dev/tags/distributed-systems), [#redis](https://daily.dev/tags/redis), [#valkey](https://daily.dev/tags/valkey) [View this post on daily.dev](https://daily.dev/posts/p99-conf-speaker-spotlight-madelyn-olson-zh968yypl) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"P99 CONF Speaker Spotlight: Madelyn Olson","url":"https://daily.dev/posts/p99-conf-speaker-spotlight-madelyn-olson-zh968yypl","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/p99-conf-speaker-spotlight-madelyn-olson-zh968yypl"},"datePublished":"2026-08-19T14:49:50.169Z","dateModified":"2026-09-14T07:55:19.843Z","description":"A conference speaker spotlight interview with Madelyn Olson, Valkey co-creator and AWS Principal Software Engineer, previewing her P99 CONF talk on the...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/c8c94a11413828ccb7ec9f046dbbaf0c?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/c8c94a11413828ccb7ec9f046dbbaf0c?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"P99 Conf","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"P99 Conf","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/f33277b331394790a6490e1b91b85bb7","url":"https://daily.dev/sources/p99conf"},"commentCount":1,"discussionUrl":"https://daily.dev/posts/p99-conf-speaker-spotlight-madelyn-olson-zh968yypl","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":5},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":1}],"keywords":"performance,backend,distributed-systems,redis,valkey","timeRequired":"PT5M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"P99 Conf","item":"https://daily.dev/sources/p99conf"},{"@type":"ListItem","position":3,"name":"P99 CONF Speaker Spotlight: Madelyn Olson"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/p99-conf-speaker-spotlight-madelyn-olson-zh968yypl","comment":[{"@type":"Comment","text":"NVMe tiering feels like a practical response to DRAM cost, especially for large AI inference caches. The important measurements will be tail latency during promotion and eviction, not just capacity gains. The B+ tree and Swiss table rewrites also sound like excellent real-world performance case studies.","datePublished":"2026-08-20T04:15:15.672Z","url":"https://daily.dev/posts/zH968yYPL#c-ucBse6MgB","author":{"@type":"Person","name":"Trevor Suna","url":"https://daily.dev/trevorsuna","image":"https://media.daily.dev/image/upload/s--dZ7gXxpp--/f_auto/v1784081551/avatars/avatar_EMoP47rpuw8DNjhp6R1b6?_a=BAMAMicg0"}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/p99-conf-speaker-spotlight-madelyn-olson-zh968yypl#faq","mainEntity":[{"@type":"Question","name":"What is data tiering in Valkey and why is it being built now?","acceptedAnswer":{"@type":"Answer","text":"Data tiering in Valkey moves some data from DRAM to NVMe storage instead of keeping everything in memory, a shift made practical because NVMe latency has dropped and bandwidth has climbed while DRAM remains the biggest cost driver in Valkey deployments. Demand comes from AI inference KV caching workloads with extreme throughput and capacity needs, plus teams seeking cost savings. AWS previously built a similar feature for ElastiCache and is now building an open source version for Valkey. Anyone tracking cost-driven storage tradeoffs can follow Valkey's tiering work as it develops on daily.dev."}},{"@type":"Question","name":"Why does Valkey support both scaling up and scaling out instead of picking one?","acceptedAnswer":{"@type":"Answer","text":"Valkey supports both because each approach has distinct failure modes: scaling up on one large node creates a bigger blast radius, longer replication times, and wasted spend when hardware doesn't match the workload, while scaling out has traditionally suffered from slow slot migration and disruptive rebalancing. Command execution remains single threaded in Valkey for now, a deliberate choice discussed alongside the metrics recommended for deciding when to scale. Teams weighing scale-up versus scale-out tradeoffs can track how Valkey's approach evolves on daily.dev."}},{"@type":"Question","name":"What internal data structure changes has Valkey made recently?","acceptedAnswer":{"@type":"Answer","text":"Valkey replaced its original ordered index, a skiplist that had been part of the codebase since the beginning, with a B+ tree, and rebuilt its main hashtable as a Swiss table. Both changes took established designs and adapted them to Valkey's specific constraints of being memory hungry while avoiding latency stalls, since the project aims to prevent cache timeouts. Developers studying database internals rewrites can follow projects like Valkey's design changes on daily.dev."}}]} ```