Cassandra Summit 2023
2023-12-12..2023-12-13
San Jose, United States
as the organisers put it
Cassandra Summit is the world’s largest gathering of Apache Cassandra community members including architects, developers, users, and industry thought leaders.
who it's for
cassandra operators, cassandra developers, site reliability engineers, data platform architects, machine learning engineers using cassandra, kubernetes platform engineers, database migration engineers
topics
Apache Cassandra 5 features; vector search with Cassandra; Cassandra performance tuning; CQL traffic mirroring for testing; Accord leaderless Paxos transactions; Cassandra backup and restore tooling; moving petabytes to Cassandra analytics; running Cassandra on Kubernetes; LLM retrieval augmented generation; GPU orchestration for LLMs on Kubernetes
| title | author | length |
|---|---|---|
| Keynote: State of Open GenAI | Jim Zemlin | 13m |
Argues that openness enables transparency, trust and attribution in generative AI, warns real risks of bias, hallucinations, privacy and security, and reports… | ||
| Keynote: Enterprise Value with Generative AI | Gourab De | 10m |
Claims that LLM development explodes experiment dimensionality and produces massive generated metadata, so teams need a system of record tying datasets, model… | ||
| Keynote : The Open Source AI Alliance | Anthony J. Annunziata | 13m |
IBM Meta Linux Foundation and 50+ partners create the AI Alliance to fund and coordinate open source and open science in AI, pooling compute, data and tooling… | ||
| Keynote: Building AI with Open Source & Hugging Face | Jeff Boudier | 14m |
Hugging Face claims it hosts over a million repositories with 400,000+ public models, reports about 10 million Transformer downloads per day, and lists 3,000+… | ||
| Keynote: Data = AI: Leading the Future of Gen AI with Apache Cassandra | Chet Kapoor | 18m |
Claims that Apache Cassandra powers real‑time, enterprise‑scale Gen AI with production-ready features: customers go live in 60 days or less, PhysicsWalla ships… | ||
| Keynote: Unlocking Corporate Data with Retrieval-Augmented Generation | Jerry Liu | 10m |
Retrieval-augmented generation handles simple fact lookup but fails on multi‑step comparison, SQL‑style filtering, and structured analytics; solution is an LLM… | ||
| Keynote: Introducing Apache Cassandra® 5! | Patrick McFadin, Scott Andreas, Ekaterina Dimitrova | 10m |
Cassandra 5 beta released with key features shipping now and 5.1 following shortly, and Cassandra 3.x moves to EOL while certifications update. Deliverables… | ||
| Improving Bad Partition Handling in Apache Cassandra | Jordan West, Cheng Wang | 25m |
Netflix runs 900+ Cassandra clusters storing 12+ PB and 12M req/s and describes a workflow that finds offending keys with profile load, stops them via an… | ||
| Operatoration Testing of Cluster | Pushkala Pattabhiraman | 14m |
DataStax runs nightly ephemeral clusters for connectivity and security checks, weekly 48‑hour soak tests on ~45‑node DSC clusters, and multi‑week runs on geo… | ||
| Bringing Apache Cassandra Closer to Kubernetes | Jake Luciani | 34m |
A Cassandra-based serverless system is built to run only on Kubernetes and is already running in production as Astra, decoupling compute and storage by using… | ||
| Embracing Multi-Cloud Cassandra at Walmart | Andrew Weaver, Patrick Lee | 33m |
Walmart runs 300 production clusters totaling multi‑petabytes on a mix of a million‑core private cloud plus Azure and GCP, 100% on VMs with most clusters on… | ||
| Sponsored Session: Cassandra on Steroids | Alex Berkoff | 27m |
Azul Platform Prime replaces OpenJDK GC and JIT with C4 continuous garbage collection, Falcon JIT and ReadyNow warm‑up profiles to eliminate stop‑the‑world… | ||
| Complex Event Processing Made Easy with Cassandra and Quine Streaming Graph | Ethan Bell, Aaron Ploetz | 34m |
Quine builds a property graph from incoming JSON logs using Cypher-like queries, runs standing queries to emit pattern matches in real time, and persists hot… | ||
| How Netflix Delivers Key-Value and Time-Series Storage at Any Scale | Joseph Lynch, Vidhya Arvind | 34m |
Netflix runs Apache Cassandra at petabyte scale using client-side LZ4 compression, chunked writes (64 KB chunks, 2 MB pages) and dynamically bucket data into… | ||
| Importance of Transactional Metadata (Or How We’re Improving Upon Gossip) | Alex Petrov | 45m |
Replaces gossip for cluster metadata with transactional cluster metadata implemented as an ordered change log of transformations, identified by monotonically… | ||
| Sponsored Session: How is Amazon Keyspaces (for Apache Cassandra) Architected to Deliver Performance at Any Scale? | Rashmi Krishnaiah Setty, Rohan Gupta | 27m |
Managed Cassandra-compatible service that stores petabyte‑scale tables with virtually unlimited rows, auto‑splits hot partitions to double throughput, and… | ||
| Designing Modern Data Platforms with Apache Cassandra and Beyond | Rahul Singh, Allison Nokes | 27m |
Presents an opinionated Data Platform Canvas and playbook that anchors Apache Cassandra (self-managed, Astra) into a full stack with Spark/EMR, Presto/Trino… | ||
| Anti-Patterns with Apache Cassandra | Praveen Kuruvadi, Nikolai Kolesnikov | 29m |
Cassandra clusters suffer two concrete anti-patterns: creating huge numbers of tables and growing oversized partitions. Table metadata consumes about ~1 MB… | ||
| SmartThings First 8 Years with Cassandra | Jeff Beck, Colleen Velo | 33m |
Eight to nine years of production Cassandra: more than 300+ nodes across over ten clusters with two global rings. Operations migrate from DataStax Enterprise… | ||
| Compaction for All! Unified Compaction Strategy in Cassandra 5.0 (CEP-26) | Lorina Poland | 39m |
Cassandra 5.0 introduces Unified Compaction Strategy (UCS) that replaces LCS/STCS with a single density-based sharding approach: SSTable density (size/token… | ||
| Memtable Pluggability and Alternate Memtables (CEP-11 & CEP-19) | Branimir Lambov, Shylaja Kokoori | 37m |
CEP-11 pluggability lets Cassandra swap memtable implementations and CEP-19 replaces the partition map with a byte‑order trie memtable using key translation… | ||
| Building a Global CyberGrid on Cassandra | Charles D. Herring | 21m |
Builds a global cyber grid that ingests hundreds of organizations and tens of terabytes per day using Kafka for ingestion, an NLP stage and graph processing… | ||
| The What and Why of Cassandra Datamodelling at Scale | Chandi Datta, Sunil Dabre | 35m |
Verizon migrates legacy DB2 mainframe workloads to Cassandra with a read-first real-time DB2 -> Cassandra replication strategy, using Cassandra timestamps to… | ||
| Sponsored Session: Cassandra One-Stop Operations with AxonOps | Hayato Shimizu, Johnny Miller | 33m |
AxonOps packages monitoring, maintenance and backups into one tool for Apache Cassandra, supporting 3.11 through 5.0 and a technical-preview cloud provisioner… | ||
| Scale Your Application to Infinity | Denson Pokta | 31m |
IPS maps client XML schemas to a single Cassandra-backed model that services use for hundreds of Intuit apps, persisting attributes in Cassandra and large… | ||
| How Uber Uses Cassandra to Power Core Applications at Scale? | Jaydeepkumar Chovatia, Gopal Mor | 33m |
Uber runs Cassandra across thousands of nodes handling millions of QPS and petabytes of data, using an internal fork with production changes for service… | ||
| Storage Attached Indexes for Cassandra (CEP-7) | Piotr Kołaczkowski | 31m |
Storage Attached Indexes (CEP-7) land in Cassandra 5 as per-sstable and per-memtable index files that attach to Memtable and SSTable so indexes share lifecycle… | ||
| The State of the Cassandra Development Community | Joshua A McKenzie | 36m |
Community participation shows a clear recovery since 2019 with an explicit count of 73 committers and 36 PMC members and steady growth in unique contributors… | ||
| Apache Cassandra Sidecar: The Companion for Use Cases in Cassandra | Francisco Guerrero Hernandez, Saranya Krishnakumar | 20m |
A reactive sidecar process built on Vert.x event loop runs alongside Cassandra, exposing REST APIs and managing one or more nodes via JMX and CQL with… | ||
| Migrating Workloads from Mainframes to Cassandra - Problems, Experience and Learnings | Seshu K Guddanti, Venkat Mopuri | 32m |
U.S. Bank moves read traffic off mainframes to Cassandra using a CQRS pattern, Kafka and Spark (structured/continuous streaming) and writes remain on the… | ||
| Securing Multi-Cluster Cassandra and GenAI Deployments | Jiong Liu | 26m |
Wiz reports over 40% of the Fortune 100 as customers and shows concrete failures: an EC2 “patient zero” exploited with Dirty Cow leads to four keys found, one… | ||
| RAG Q and A with Bedrock and Astra - Who died in Romeo and Juliet? | Kirsten Hunter | 30m |
Builds a working RAG pipeline that loads 321 lines (one line per document) from Act 5 Scene 3 into a vector store and answers “who died” using embeddings and… | ||
| Apache Cassandra Analytics: A Recipe to Move Petabytes of Data | Dinesh A. Joshi | 31m |
Two open source components enable bulk moves of petabytes by streaming SSTables instead of row I O the Cassandra Analytics Library runs on Spark to deserialize… | ||
| OpenTelemetry and Apache Cassandra | Yuki Morishita | 23m |
Integration of OpenTelemetry protocol into Apache Cassandra enables exporting traces, metrics and logs to Datadog, Honeycomb and CloudWatch from a 3-node… | ||
| 70 TB Servers – How and Why We Use Cassandra | Nathan Jackels | 28m |
Claims running up to ~80 TB per physical server using Cassandra with production customers and it works: achieves dense 5×16 TB instance layouts on Dell/HP… | ||
| Astacus: Cloud Storage Backup and Restore Tool for Apache Cassandra | Dmitry Potepalov | 23m |
Astacus is a Python tool that coordinates per-node backup and restore for Apache Cassandra via a REST-like coordinator API and HTTP inter-node calls, using… | ||
| Code That Can't Fail | Loren Sands-Ramshaw | 18m |
Explains how durable execution makes a function guaranteed to resume after crashes, redeploys or long sleeps by persisting each step; implementation uses… | ||
| Reducing Client Latency and Timeouts When Running Cassandra on Public Cloud | German Eichberger | 27m |
Public cloud maintenance and VM freezes produce multi‑second tail latencies (observed ~8–30s) that make local-one reads and local quorum fail or return stale… | ||
| Keynote: How LLMs Have Changed the Way Organizations Do AI | Robert Nishihara | 12m |
Notes on LLM adoption: organizations move from a fast build-and-learn phase into a scale-and-optimize phase, shifting priorities from quality to cost, latency… | ||
| Keynote: Enabling AI with Cloud Native | Jorge Castro | 10m |
Highlights hardware and orchestration work that makes generative AI run on cloud native stacks: GPU/TPU/FPGA slicing via KEPs and dynamic resource allocation… | ||
| Keynote: Build Accelerated LLM Applications with Open Source Tools from NVIDIA | Ankit Patel | 12m |
NVIDIA claims 500 open-source projects and says over 100 million RTX PCs exist to run them, then demos concrete tools: RAPIDS accelerator runs existing pandas… | ||
| Keynote: Accelerating AI/ML Development with Open Source Software | Brian E. Granger | 16m |
AWS claims two concrete strategies: product integrations that package open source for secure, scalable use, and direct contributions to projects. Examples… | ||
| Keynote: Scaling Open Source Large Language Models | Charles Srisuwananukorn | 10m |
Together builds the Together Inference API to serve 100+ open source LLMs and shows it works but not without surprises. Model startup time is dominated by… | ||
| Keynote: Teaching LLMs New Knowledge | Sharon Zhou | 05m |
Lamini demonstrates concrete ways to inject new domain knowledge into open LLMs: fine-tuning and pretraining combined with prompting and RAG, plus tooling that… | ||
| Keynote: What's New in AI | Eve Phillips | 12m |
Overview of concrete advances: attention Transformers (2017) enable longer-context models, hardware like TPUs (2015) and GPUs scale training to models with… | ||
| Keynote: Next-Gen AI Applications via Multi-Agent Conversation | Victor Dibia | 10m |
AutoGen is presented as an open source multi-agent conversation framework that composes specialized LLM agents (including a user proxy agent, an assistant… | ||
| OLAP on Your Cassandra Data with Arrow, Flight SQL, ADBC, and DuckDB | Philip Moore | 30m |
Demonstrates converting Cassandra SSTable snapshots to Parquet and running OLAP via an Arrow/Flight SQL stack with DuckDB on Graviton, deployed in Kubernetes… | ||
| Fast and Flexible JSON Retrieval in Cassandra Using SAI | Aaron Morton | 29m |
Method stores the JSON blob and creates a “super‑shredded” row per document that indexes paths and typed values so CQL + SAI can answer filters like existence… | ||
| Observability and Debuggability in Cassandra | David Paulk | 33m |
Datadog SREs collect JMX, GC logs and request traces to tie latency spikes to resource signals and to validate fixes: long GC pauses from G1 humongous objects… | ||
| Apache Cassandra: What’s New and What’s Next | Scott Andreas | 30m |
Apache Cassandra now adds concrete features: Accord leaderless Paxos implements cross-partition transactions with strict serializability and usually a… | ||
| Cassandra on ACID: This Changes Everything | Patrick McFadin | 34m |
Cassandra adds ACID transactions via Accord and the CEP-15 branch, promising strict serializable guarantees with a new CQL block: begin transaction / commit… | ||
| Popular Apache Cassandra Problems & How to Fix 'Em | Michael Semb Wever | 37m |
Eight concrete fixes and measurable checks for Cassandra operations: run 10,000 IOPS soak test on each node to validate SSD controllers, use G1 and keep heap… | ||
| Guerilla Tactics for Building Scalable E-Commerce Services with Apache Cassandra, Apache Pulsar, and Vector Search | Aaron Ploetz | 28m |
A working e-commerce prototype uses Java Spring Boot, Spring Data Cassandra on AstraDB, Apache Pulsar for order events, and a Python loader with Hugging Face +… | ||
| Scraping Your Cassandra Metrics - the Right Way, and the Dirty Way | Hayato Shimizu | 28m |
Cassandra metrics generate hundreds of thousands of MBeans and scraping them causes big CPU spikes, so AxonOps builds a Java agent (Byte Buddy) that… | ||
| Operating Cassandra in Kubernetes with Temporal | Christian Mauduit, Steven Seidman | 26m |
Datadog runs Cassandra across hundreds of clusters and thousands of nodes on Kubernetes and replaces ad-hoc scripts with Temporal workflows that orchestrate… | ||
| Cassandra Performance Tuning Like You've Been Doing It for Ten Years | Jon Haddad | 33m |
Explains a repeatable observe‑orient‑decide‑act methodology plus Brendan Gregg’s USE checks (utilization, saturation, errors) and concrete tools… | ||
| Database Native Support for CDC Streaming Pipeline | Neha Maheshwari, Vinit Gupta | 24m |
Most Cassandra CDC pipelines read commit logs, deduplicate across replicas (replication factor 3 producing up to nine replicas), reconstruct full before/after… | ||
| Testing in Production – Mirroring CQL Traffic to a Shadow Cassandra Deployment | Piotr Kołaczkowski, Jakub Żytka | 36m |
A lightweight Rust TCP proxy duplicates CQL byte streams to per-node shadow clusters in Kubernetes, with online reconfiguration, per-session and global… | ||
| Making Cassandra Easier for AI Developers | Jeffrey Carpenter | 33m |
Cassandra becomes usable for AI by adding a document-layer HTTP API and vector features: a new Json API plus Storage Attached Indexes let Cassandra act like a… | ||
| Stories of Cassandra in Adobe Audience Manager | Serban Teodorescu, Stefan Gusa | 36m |
Adobe Audience Manager runs 34 Cassandra clusters across eight AWS regions on 800+ instances holding 260 TB and serving roughly 313 billion Cassandra requests… | ||
| How Cassandra 4.X Help in Cutting Cost Mission | Robert Czupiol | 26m |
Self-hosted cluster shrinks from 42 nodes to a footprint that drops Cassandra from the top bill to 20th position, achieving about 69% TCO reduction. Savings… | ||
| Building a ChatGPT Data Pipeline with RisingWave Stream Processor and Cassandra Vector Search | Mary Grygleski, Karin Wolok | 32m |
Shows a working pipeline that ingests Kafka Twitter feeds into RisingWave stream processor, runs SQL transforms and incremental materialized views, chunks text… | ||
| Migrating Applications Between Cassandra Clusters with No Downtime | Jeremy Hanna, Wei Deng | 32m |
Open-source Apache v2 tooling provides a stateless CQL proxy (Go, Docker) that does synchronous dual-write to origin and target clusters and only returns… | ||
| Sponsored Session: Building a Production Open Source RAG Application using LlamaIndex and Cassandra | Patrick McFadin, Yi Ding | 31m |
Datastax shows a working knowledge-retrieval chatbot built with LlamaIndex and Astra as the vector store, integrated with Intercom and Slack, and released as… | ||
| Sponsored Session: AI-Powered Semantic Search with Cassandra: Today and Tomorrow | Murilo Miranda, Bassam Chahine | 28m |
Shows how to build semantic search with fastText embeddings, storing 32 paragraph chunks and handling an eight‑minute model load; with Cassandra 4 + OpenSearch… | ||
| Deciphering Cassandra’s Prophecy: Preventing the Fall of Your Application | Isaac Reath, Lindsey Zurovchak | 25m |
Bloomberg runs Cassandra as a managed service at 2.5 PB, 3,000 nodes, 250 clusters and shares concrete failures and fixes: mixed TTLs plus… | ||
| Enhancing Apache Cassandra Drivers with Vector Support | Bret McGuire | 32m |
Three drivers (Java, Python, Node.js) now include native vector support for Astra Vector and DSE 7.0 preview with planned compatibility for Apache Cassandra… | ||
| Introduction to NoSQLBench | Jonathan Shook | 33m |
NoSQLBench is a multiprotocol load generator that uses procedural data generation and driver adapters to run reproducible cycles with O(1) sampling for large… | ||
| Unleash the Power of Apache Cassandra: Scalable Recommendation Systems, Real-time Fraud Detection | Himanshu Jindal, Ekta Sachan | 26m |
Shows Apache Cassandra (via Amazon Keyspaces) as the primary operational store with concrete demos: a SageMaker pipeline that reads ~20,000 transactions… | ||
| Intel and Open Source: Bringing AI Everywhere | Arun Gupta | 24m |
Intel claims broad, practical work to make AI ubiquitous: contributing to 300+ open-source projects, remaining the only CPU maintainer of PyTorch and leading… | ||
| Bring Cassandra to the GenAI Crowd: Meet the Low-Friction CassIO Library | Stefano Lottini | 21m |
CassIO is a Python library that uses Cassandra 5’s CEP-30 vector support to let GenAI code create and query production-ready vector tables with one-line vector… | ||
| Navigating Green Waters: A Journey with Cassandra | Praveen Viswanath, Timothy Tran | 32m |
Maritime IoT platform that runs Cassandra on both ships and cloud handles 3,000–5,000 datapoints per 30s from ~450–1,600+ vessels, drove 230,000 tCO2 and $38M… | ||
| Leveraging Cassandra to Provide Generative AI That Powers Enterprise Analytics | Alan Ho | 19m |
Skypoint builds an enterprise AI lakehouse that combines Databricks structured tables, vector retrieval over Astra DB, LangChain agents and fine‑tuned LLMs to… | ||
| nothing matches | ||