<!DOCTYPE html><html lang="en"><head><meta http-equiv="Content-Type" content="text/html charset=UTF-8"><meta charset="UTF-8"><meta name="viewport" content="width=device-width"><meta name="x-apple-disable-message-reformatting"><title>TLDR Data</title><meta name="color-scheme" content="light dark"><meta name="supported-color-schemes" content="light dark"><style type="text/css">
:root {
color-scheme: light dark; supported-color-schemes: light dark;
}
*,
*:after,
*:before {
-webkit-box-sizing: border-box; -moz-box-sizing: border-box; box-sizing: border-box;
}
* {
-ms-text-size-adjust: 100%; -webkit-text-size-adjust: 100%;
}
html,
body,
.document {
width: 100% !important; height: 100% !important; margin: 0; padding: 0;
}
body {
-webkit-font-smoothing: antialiased; -moz-osx-font-smoothing: grayscale; text-rendering: optimizeLegibility;
}
div[style*="margin: 16px 0"] {
margin: 0 !important;
}
table,
td {
mso-table-lspace: 0pt; mso-table-rspace: 0pt;
}
table {
border-spacing: 0; border-collapse: collapse; table-layout: fixed; margin: 0 auto;
}
img {
-ms-interpolation-mode: bicubic; max-width: 100%; border: 0;
}
*[x-apple-data-detectors] {
color: inherit !important; text-decoration: none !important;
}
.x-gmail-data-detectors,
.x-gmail-data-detectors *,
.aBn {
border-bottom: 0 !important; cursor: default !important;
}
.btn {
-webkit-transition: all 200ms ease; transition: all 200ms ease;
}
.btn:hover {
background-color: #f67575; border-color: #f67575;
}
* {
font-family: Arial, Helvetica, sans-serif; font-size: 18px;
}
@media screen and (max-width: 600px) {
.container {
width: 100%; margin: auto;
}
.stack {
display: block!important; width: 100%!important; max-width: 100%!important;
}
.btn {
display: block; width: 100%; text-align: center;
}
}
body,
p,
td,
tr,
.body,
table,
h1,
h2,
h3,
h4,
h5,
h6,
div,
span {
background-color: #FEFEFE !important; color: #010101 !important;
}
@media (prefers-color-scheme: dark) {
body,
p,
td,
tr,
.body,
table,
h1,
h2,
h3,
h4,
h5,
h6,
div,
span {
background-color: #27292D !important; color: #FEFEFE !important;
}
}
a {
color: inherit !important; text-decoration: underline !important;
}
</style><!--[if mso | ie]>
<style type="text/css">
a {
background-color: #FEFEFE !important; color: #010101 !important;
}
@media (prefers-color-scheme: dark) {
a {
background-color: #27292D !important; color: #FEFEFE !important;
}
}
</style>
<![endif]--></head><body class="">
<div style="display: none; max-height: 0px; overflow: hidden;">Meta's Zoomer is an automated, end-to-end platform for debugging and optimizing AI workloads. It has slashed training times </div>
<div style="display: none; max-height: 0px; overflow: hidden;">
<br>
</div>
<table align="center" class="document"><tbody><tr><td valign="top">
<table align="center" border="0" cellpadding="0" cellspacing="0" class="container" width="600"><tbody><tr class="inner-body"><td>
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr class="header"><td bgcolor="" class="container">
<table width="100%"><tbody><tr><td class="container">
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" style="margin-top: 0px;" width="100%"><tbody><tr><td style="padding: 0px;">
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div style="text-align: center;">
<span style="margin-right: 0px;"><a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Ftldr.tech%2Fdata%3Futm_source=tldrdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/TJdeaDuDRCMGSs4S4LTCXFcbPb2V9UfU_hWdUuG9vPk=432" rel="noopener noreferrer" target="_blank"><span>Sign Up</span></a>
|<span style="margin-right: 2px; margin-left: 2px;"><a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fadvertise.tldr.tech%3Futm_source=tldrdata%26utm_medium=newsletter%26utm_campaign=advertisetopnav/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/2-9lhGnOuUsxB9TiWhxjr6_QiqS-NBl_kuKNHb7dQXo=432" rel="noopener noreferrer" target="_blank"><span>Advertise</span></a></span>|<span style="margin-left: 2px;"><a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fa.tldrnewsletter.com%2Fweb-version%3Fep=1%26lc=1670a604-84b7-11f0-bcf5-55fc1d40139c%26p=0c08c07e-c900-11f0-b75c-63b81888de10%26pt=campaign%26t=1763982395%26s=4a2bafbbcca8018b0bac785deabdacaaaa11d5d584624cc822f87a2a281f7caa/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/1p9keksoe-T8eNJxOGRv9SVrAUnlmLRfs5s7uQRv5Bo=432"><span>View Online</span></a></span>
<br>
</span></div>
</td></tr></tbody></table>
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="text-align: center;"><span data-darkreader-inline-color="" style="--darkreader-inline-color:#3db3ff; color: rgb(51, 175, 255) !important; font-size: 30px;">T</span><span style="font-size: 30px;"><span data-darkreader-inline-color="" style="color: rgb(232, 192, 96) !important; --darkreader-inline-color:#e8c163; font-size:30px;">L</span><span data-darkreader-inline-color="" style="color: rgb(101, 195, 173) !important; --darkreader-inline-color:#6ec7b2; font-size:30px;">D</span></span><span data-darkreader-inline-color="" style="--darkreader-inline-color:#dd6e6e; color: rgb(220, 107, 107) !important; font-size: 30px;">R</span>
<br>
</td></tr></tbody></table>
<br>
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody></tbody></table>
<table style="table-layout: fixed; width:100%;" width="100%"><tbody><tr><td style="padding:0;border-collapse:collapse;border-spacing:0;margin:0;">
<div style="text-align: center;">
<h1><strong>TLDR Data <span id="date">2025-11-24</span></strong></h1>
</div>
</td></tr></tbody></table>
<table style="table-layout: fixed; width:100%;" width="100%"><tbody></tbody></table>
</td></tr></tbody></table>
</td></tr></tbody></table>
</td></tr>
<tr bgcolor=""><td class="container">
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td style="padding: 0px;">
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding-top: 0px; padding-bottom: 0px;">
<div class="text-block">
<div style="text-align: center;"><span style="font-size: 36px;">📱</span></div></div>
</td></tr></tbody></table>
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding-top: 0px; padding-bottom: 0px;">
<div class="text-block">
<div style="text-align: center;">
<h1><strong>Deep Dives</strong></h1>
</div>
</div>
</td></tr></tbody></table>
<table style="table-layout: fixed; width: 100%;" width="100%"><tbody><tr><td style="padding:0;border-collapse:collapse;border-spacing:0;margin:0;" valign="top">
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fengineering.fb.com%2F2025%2F11%2F21%2Fdata-infrastructure%2Fzoomer-powering-ai-performance-meta-intelligent-debugging-optimization%2F%3Futm_source=tldrdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/vctPuYibtSZlkqRJb3ma1MxIKQHa3Hx8TPN0oS01dNw=432">
<span>
<strong>Zoomer: Powering AI Performance at Meta's Scale Through Intelligent Debugging and Optimization (8 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
Meta's Zoomer is an automated, end-to-end platform for debugging and optimizing AI workloads. It has slashed training times (75% reduction for Ads relevance models, yielding 78% less power use) and boosted inference QPS by ~20% with a one-line code change that fixed inefficient memory copy, and enabled 25-30% speedups on massive 32k-64k GPU jobs, reducing environmental impact from trillions of daily inferences.
</span>
</span>
</div>
</td></tr></tbody></table>
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fwww.linkedin.com%2Fblog%2Fengineering%2Finfrastructure%2Fevolution-of-the-venice-ingestion-pipeline%3Futm_source=tldrdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/8Dulut4x07eHMux5Z72NY1ZQqSzxCDrZcCMWmZx0DWE=432">
<span>
<strong>The Evolution of the Venice Ingestion Pipeline (15 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
LinkedIn's Venice ingestion pipeline is an open-source derived data storage system that serves as the backbone for online AI applications like People You May Know, feed recommendations, ads, and notifications. Launched in 2016, Venice now powers over 2,600 production stores, optimized for I/O, memory, and CPU efficiency across hybrid and active-active setups. It handles over 175 million key lookups per second and 230 million writes per second, all while maintaining a write latency SLA of under 10 minutes.
</span>
</span>
</div>
</td></tr></tbody></table>
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fhackernoon.com%2Fhow-search-engines-actually-answer-your-questions%3Futm_source=tldrdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/OoQV1RW5qlMZO-VkRD2YMw5zxEzIrUXdLL4X4cFVuHA=432">
<span>
<strong>How Search Engines Actually Answer Your Questions (6 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
Modern search Q&A systems leverage a hybrid architecture that combines knowledge graph-based QA (KBQA) for precision on structured facts with DeepQA and machine reading comprehension for coverage across unstructured web data. Production stacks employ fusion layers (meta-rankers evaluating relevance, freshness, and trust) while neural models enhance robustness via techniques like R-Drop and data augmentation. Key advances include generative readers for nuanced answers, HTML-aware encoding, structure-aware pretraining, and explicit evidence highlighting.
</span>
</span>
</div>
</td></tr></tbody></table>
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fthinhdanggroup.github.io%2F%2Fsqlite-serverless-backend%2F%3Futm_source=tldrdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/4nvZYxi8_hkKEGKZS3Tixw5dlVyPmejXsOYePawXHs8=432">
<span>
<strong>How SQLite Is Powering the New Generation of Serverless Backends (7 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
Modern serverless architectures are leveraging SQLite as a high-performance, globally distributed storage engine by wrapping it with platforms like Turso/libSQL, Cloudflare D1, and LiteFS. These solutions extend SQLite's durability and embedded nature by adding replication, edge distribution, and session-based consistency, enabling local, low-latency reads and synchronized writes via a primary node, all while minimizing operational complexity and cost. The shift enables “scale-to-zero” economics and simplifies deployment for read-heavy, multi-region backends, although single-writer replication models mean high-volume multi-region writes or conflict-free sync require more specialized data platforms. For classic SaaS and web app workloads: SQLite, when paired with the right distributed system wrapper, is now a practical, efficient backbone for serverless, edge-native data platforms.
</span>
</span>
</div>
</td></tr></tbody></table>
</td></tr></tbody></table>
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding-top: 0px; padding-bottom: 0px;">
<div class="text-block">
<div style="text-align: center;"><span style="font-size: 36px;">🚀</span></div>
</div>
</td></tr></tbody></table>
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding-top: 0px; padding-bottom: 0px;">
<div class="text-block">
<div style="text-align: center;">
<h1><strong>Opinions & Advice</strong></h1>
</div>
</div>
</td></tr></tbody></table>
<table style="table-layout: fixed; width: 100%;" width="100%"><tbody><tr><td style="padding:0;border-collapse:collapse;border-spacing:0;margin:0;" valign="top">
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fwww.dataengineeringweekly.com%2Fp%2Fthe-dark-data-tax-how-hoarding-is%3Futm_source=tldrdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/J7RY42fRyA-y1p8TvT5ioMT4o4ik2SskQ7QwyqrEB6Q=432">
<span>
<strong>The Dark Data Tax: How Hoarding is Poisoning Your AI (7 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
Enterprises now store 2.5x more data than in 2019, but only a fraction is actively analyzed, resulting in “data obesity”, where operational and cognitive costs far outweigh storage, which accounts for just 8% of ownership costs. Dark data (over 68% of stored data) degrades analytics and LLM performance, increasing compliance risks and engineering overhead. This post proposes a “Data Sustainability Index” metric based on how much compute data generates relative to the overall cost and complexity of the system as a key indicator to help organizations steer away from data hoarding and towards maximal data utility.
</span>
</span>
</div>
</td></tr></tbody></table>
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Flinks.tldrnewsletter.com%2FOuiIiU/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/65X0ijPGdGcqkHPCMtKLC3I_h5JfK1ITCSodMub-hbM=432">
<span>
<strong>Metadata: How Data About Your Data is Optimal for AI (12 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
Metadata is data about data that describes, categorizes, and tracks information through tags, lineage, and standardized glossaries, transforming raw datasets into rich, contextual sources that power accurate AI models. It serves as the foundational framework for AI pipelines by enabling efficiency through experiment tracking and schema standardization, ensuring reproducibility, and automating smarter feature selection and insights.
</span>
</span>
</div>
</td></tr></tbody></table>
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fwww.morling.dev%2Fblog%2Fbuilding-durable-execution-engine-with-sqlite%2F%3Futm_source=tldrdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/Qf7qwUuP8kmCXlzKZaSFN5gSjg7s2BPgYM1vdKZPz98=432">
<span>
<strong>Building a Durable Execution Engine With SQLite (12 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
Persistasaurus, a lightweight Durable Execution (DE) engine in Java using SQLite for persistence, ensures workflows resume from interruptions without re-executing completed steps by logging intents and outcomes for replay or resumption. Benefits include avoiding duplicates, capturing non-determinism, and enabling long-running processes to complete reliably.
</span>
</span>
</div>
</td></tr></tbody></table>
</td></tr></tbody></table>
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding-top: 0px; padding-bottom: 0px;">
<div class="text-block">
<div style="text-align: center;"><span style="font-size: 36px;">💻</span></div>
</div>
</td></tr></tbody></table>
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding-top: 0px; padding-bottom: 0px;">
<div class="text-block">
<div style="text-align: center;">
<h1><strong>Launches & Tools</strong></h1>
</div>
</div>
</td></tr></tbody></table>
<table style="table-layout: fixed; width: 100%;" width="100%"><tbody><tr><td style="padding:0;border-collapse:collapse;border-spacing:0;margin:0;" valign="top">
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Flanguages.oup.com%2Fproducts%2Fon-demand-webinar-cracking-the-code-building-nlp-for-complex-languages%2F%3Futm_source=on-demand%2Bwebinar%2Bnlp%2B24112025%26utm_medium=tldr%2Bdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/mrNV4hUCJbL91ppVdwY79CL6-Q972q8qe65DtFNCERU=432">
<span>
<strong>The evolution of AI translation tools as told by Oxford Languages (Sponsor)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
Learn how Finn language tech Kielikone Oy built domain-specific datasets to meet the linguistic demands of Finnish users. This <a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Flanguages.oup.com%2Fproducts%2Fon-demand-webinar-cracking-the-code-building-nlp-for-complex-languages%2F%3Futm_source=on-demand%2Bwebinar%2Bnlp%2B24112025%26utm_medium=tldr%2Bdata/2/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/EvABlq4ayt5VOZUj6eDG5Njn2sZz4ulWX6pjOmzcISA=432" rel="noopener noreferrer nofollow" target="_blank"><span>webinar</span></a> explains what happened when even the best NLP models didn't quite meet the morphological needs of a language with 15 grammatical cases. <a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Flanguages.oup.com%2Fproducts%2Fon-demand-webinar-cracking-the-code-building-nlp-for-complex-languages%2F%3Futm_source=on-demand%2Bwebinar%2Bnlp%2B24112025%26utm_medium=tldr%2Bdata/3/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/qB54MuryEDHK8BOI6d1SYaOCidPDucMSiy08kOBndP0=432" rel="noopener noreferrer nofollow" target="_blank"><span>Watch the webinar</span></a>
</span>
</span>
</div>
</td></tr></tbody></table>
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fduckdb.org%2F2025%2F11%2F19%2Fencryption-in-duckdb%3Futm_source=tldrdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/II20C_l1EOvrjOb06gKwt06HFBMRLjY1RAkuJFWnIdY=432">
<span>
<strong>Data-at-Rest Encryption in DuckDB (22 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
DuckDB 1.4 introduces built-in data-at-rest encryption, allowing transparent AES-based protection of database files stored on disk. This feature simplifies secure analytics by encrypting the entire database (including WAL and temps), reducing threats from lost media or access breaches. It supports compliance, enables key-based read-only sharing, and maintains DuckDB's edge in portable, high-performance OLAP.
</span>
</span>
</div>
</td></tr></tbody></table>
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fjack-vanlightly.com%2Fblog%2F2025%2F11%2F19%2Fhave-your-iceberg-cubed-not-sorted-meet-qbeast-the-otree-spatial-index%3Futm_source=tldrdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/vmPrhIEu9VymmgzFhhoJLDcapG2huZlCI1vvBikqBTE=432">
<span>
<strong>Have your Iceberg Cubed, Not Sorted: Meet Qbeast, the OTree Spatial Index (11 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
OTree is a lightweight, adaptive, multidimensional spatial index for Apache Iceberg and Delta Lake that dynamically shapes data layout using hypercube partitioning rather than static partitioning or sort order. It minimizes data drift, ensures efficient clustering, and adapts to evolving data distributions by recursively subdividing data space based on indexed column distributions. This approach aims to strike a balance between the database's classic B-tree clustered index and Open Table Format's upfront partitioning and sorting to maintain optimal data locality and decouple layout optimization from the query engines.
</span>
</span>
</div>
</td></tr></tbody></table>
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fgrafana.com%2Fblog%2F2025%2F11%2F03%2Fgrafana-mimir-3-0-release-all-the-latest-updates%2F%3Futm_source=tldrdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/bQJimCxkm99wJTE_uWaf_LmN6SVQfAyk3HRGw3BI8W4=432">
<span>
<strong>Grafana Mimir 3.0 Release: Performance Improvements, a New Query Engine, and More (5 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
Grafana Mimir 3.0 introduces a fully decoupled architecture that separates ingestion from the query engine. Ingesters feed into Kafka and object storage, while a new streaming query engine handles data with dramatically lower memory usage (up to 92% less) and increased stability under load. Large deployments already report up to 15% lower resource usage with better reliability, making Mimir 3.0 a strong choice for large-scale metrics backends.
</span>
</span>
</div>
</td></tr></tbody></table>
</td></tr></tbody></table>
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding-top: 0px; padding-bottom: 0px;">
<div class="text-block">
<div style="text-align: center;"><span style="font-size: 36px;">🎁</span></div></div>
</td></tr></tbody></table>
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding-top: 0px; padding-bottom: 0px;">
<div class="text-block">
<div style="text-align: center;"><strong><h1>Miscellaneous</h1></strong></div>
</div>
</td></tr></tbody></table>
<table bgcolor="" style="table-layout: fixed; width: 100%;" width="100%"><tbody><tr><td style="padding:0;border-collapse:collapse;border-spacing:0;margin:0;" valign="top">
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fwww.infoq.com%2Fnews%2F2025%2F11%2Fnetflix-data-deletion%2F%3Futm_source=tldrdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/vgv39mOY1p50r9uU0Hdv9_Q15pcUxOO2b089QTzPql8=432">
<span>
<strong>Netflix Tackles Data Deletion at Scale with Centralized Platform Architecture (3 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
Netflix has engineered a centralized data deletion platform that orchestrates secure deletion across 1,300 heterogeneous datasets and multiple storage engines, successfully processing 76.8 billion row deletions with zero data loss incidents. The architecture prioritizes durability, availability, and correctness through asynchronous processing, comprehensive auditing, and robust safeguards like backpressure, rate limiting, and recovery services.
</span>
</span>
</div>
</td></tr></tbody></table>
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Flinks.tldrnewsletter.com%2FZjz8lU/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/l2bOAlHecl78eeYSXSSCy7wIGN_GeunHydL5rYJBiHI=432">
<span>
<strong>From LLMs to Time Series — The Next Wave of AI Foundation Models (3 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
Amazon's Chronos-2 is a Time Series Foundation Model (TSFM) with open weights that enables plug-and-play forecasting across complex datasets without feature engineering or retraining. Trained on millions of time series, Chronos-2 accurately predicts demand influenced by seasonality, weather, and events like Black Friday, delivering production-grade forecasts in minutes.
</span>
</span>
</div>
</td></tr></tbody></table>
</td></tr></tbody></table>
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding-top: 0px; padding-bottom: 0px;">
<div class="text-block">
<div style="text-align: center;"><span style="font-size: 36px;">⚡</span></div></div>
</td></tr></tbody></table>
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding-top: 0px; padding-bottom: 0px;">
<div class="text-block">
<div style="text-align: center;">
<h1><strong>Quick Links</strong></h1>
</div>
</div>
</td></tr></tbody></table>
<table bgcolor="" style="table-layout: fixed; width: 100%;" width="100%"><tbody><tr><td style="padding:0;border-collapse:collapse;border-spacing:0;margin:0;" valign="top">
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fwww.ssp.sh%2Fblog%2Fcost-analyzer-aws-gcp%2F%3Futm_source=tldrdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/4x9Y0k5jWTws-mqfR3GNXzAIJaWcGMvuQQfTqrNdIKY=432">
<span>
<strong>Multi-Cloud Cost Analytics: From Cost-Export to Parquet to Rill (10 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
A cloud FinOps project based on an open-source stack demonstrates how to unify AWS, GCP, and other SaaS cost data with revenue data into daily monitoring dashboards.
</span>
</span>
</div>
</td></tr></tbody></table>
<table align="center" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block">
<span>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fbrooker.co.za%2Fblog%2F2025%2F11%2F20%2Fwhat-now.html%3Futm_source=tldrdata/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/wSHpYovxkrMx_i-l0qVFaAx1qrQpcFqCNWQ8ba7j39o=432">
<span>
<strong>What Now? Handling Errors in Large Systems (3 minute read)</strong>
</span>
</a>
<br>
<br>
<span style="font-family: "Helvetica Neue", Helvetica, Arial, Verdana, sans-serif;">
The Cloudflare outage, sparked by a single .unwrap(), showed that whether to crash or degrade on error isn't a local code decision, but a global system property driven by failure correlation, architectural tolerance for crashes, and the safety of continuing with partial state.
</span>
</span>
</div>
</td></tr></tbody></table>
</td></tr></tbody></table>
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td align="left" style="word-break: break-word; vertical-align: top; padding: 5px 10px;">
<p style="padding: 0; margin: 0; font-size: 22px; color: #000000; line-height: 1.6; font-weight: bold;">
Want to advertise in TLDR? 📰
</p>
<div class="text-block" style="margin-top: 10px;">
If your company is interested in reaching an audience of data engineering professionals and decision makers, you may want to <a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fadvertise.tldr.tech%2F%3Futm_source=tldrdata%26utm_medium=newsletter%26utm_campaign=advertisecta/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/4LKQ98tPrA4PjiXZxgU_ER_HC72gmiTj394vNFO19Ys=432"><strong><span>advertise with us</span></strong></a>.
</div>
<br>
<!-- New "Want to work at TLDR?" section -->
<p style="padding: 0; margin: 0; font-size: 22px; color: #000000; line-height: 1.6; font-weight: bold;">
Want to work at TLDR? 💼
</p>
<div class="text-block" style="margin-top: 10px;">
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fjobs.ashbyhq.com%2Ftldr.tech/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/nZWMKbZ8CEyuP2n33UrD29cqE79eIZsNhxHgopnM8j4=432" rel="noopener noreferrer" style="color: #0000EE; text-decoration: underline;" target="_blank"><strong>Apply here</strong></a> or send a friend's resume to <a href="mailto:jobs@tldr.tech" style="color: #0000EE; text-decoration: underline;">jobs@tldr.tech</a> and get $1k if we hire them!
</div>
<br>
<div class="text-block">
If you have any comments or feedback, just respond to this email!
<br>
<br> Thanks for reading,
<br>
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fwww.linkedin.com%2Fin%2Fjoelvanveluwen%2F/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/ud09FYkejcC044HbKfCM67WiuC6beFc9vM6BnrEpWM4=432"><span>Joel Van Veluwen</span></a>, <a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fwww.linkedin.com%2Fin%2Fjennytzurueyching%2F/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/yRR7a01Pltreah3QbISWtczCGZooMIwN6WIpOCa1S1Y=432"><span>Tzu-Ruey Ching</span></a> & <a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fwww.linkedin.com%2Fin%2Fremi-turpaud%2F/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/Qw6Au7mafQ-IBeUK4lHhM9IVZSkfoIDgZ-Til2Vbszc=432"><span>Remi Turpaud</span></a>
<br>
<br>
</div>
<br>
</td></tr></tbody></table>
<table align="center" bgcolor="" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td class="container" style="padding: 15px 15px;">
<div class="text-block" id="testing-id">
<a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Ftldr.tech%2Fdata%2Fmanage%3Femail=silk.theater.56%2540fwdnl.com/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/yxE8N90lhmKePJEPFKBu_p4onCqZrOHNHgFLsmJavTg=432">Manage your subscriptions</a> to our other newsletters on tech, startups, and programming. Or if TLDR Data isn't for you, please <a href="https://tracking.tldrnewsletter.com/CL0/https:%2F%2Fa.tldrnewsletter.com%2Funsubscribe%3Fep=1%26l=037ede50-92cc-11ee-b0f2-b761aa2217ad%26lc=1670a604-84b7-11f0-bcf5-55fc1d40139c%26p=0c08c07e-c900-11f0-b75c-63b81888de10%26pt=campaign%26pv=4%26spa=1763982059%26t=1763982395%26s=88cc27a32edd2de1f5b68cca1c8278e920b38c8345a421785b1bbab931462712/1/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/oVhJ-n0MFuNjkC3jguq8hFlPsfjecqEg6HbFp2ITs9I=432">unsubscribe</a>.
<br>
</div>
</td></tr></tbody></table>
</td></tr></tbody></table>
</td></tr></tbody></table>
</td></tr></tbody></table>
</td></tr></tbody></table>
<img alt="" src="http://tracking.tldrnewsletter.com/CI0/0100019ab58b49ff-a35d2b33-8b1b-4266-a7e3-70209240245e-000000/40Mluc-9l5PLiNcD1R5WwogkFjarHqakMMIPYRp-Fhk=432" style="display: none; width: 1px; height: 1px;">
</body></html>