| Redis Scaling Horror Stories Episode 8: Zombie Nodes
Your cluster is overprovisioned, and you’re paying for every zomb… |
@dragonflydbio |
Company |
Original |
2026-10-30 |
126 |
0 |
0 |
0 |
0 |
0 |
| ShareChat started with Dragonfly Community Edition on Kubernetes, watched the claims hold up on their own workloads, an… |
@dragonflydbio |
Company |
Original |
2026-10-30 |
200 |
2 |
0 |
0 |
0 |
0 |
| Three years of Dragonfly in under a minute: full Redis ecosystem compatibility, better performance, predictable memory,… |
@dragonflydbio |
Company |
Original |
2026-10-29 |
177 |
2 |
0 |
0 |
0 |
0 |
| Redis Scaling Horror Stories Episode 7: The Reshard That Woke the Dead
You reshard because you’re overloaded. AWS says… |
@dragonflydbio |
Company |
Original |
2026-10-29 |
103 |
0 |
0 |
0 |
0 |
0 |
| Redis Scaling Horror Stories Episode 6: The Frankenstein
You fanned out to every shard, merged in the app, and called … |
@dragonflydbio |
Company |
Original |
2026-10-28 |
98 |
1 |
0 |
0 |
0 |
0 |
| Tomorrow at 9 AM PT, we'll show how Instacart went from 348 cache nodes to 100 and how Hugging Face scaled to 3 million… |
@dragonflydbio |
Company |
Original |
2026-10-28 |
93 |
0 |
0 |
0 |
0 |
0 |
| Redis Scaling Horror Stories Episode 5: The Failover Stampede
Automatic failover worked. Then every client reconnected … |
@dragonflydbio |
Company |
Original |
2026-10-27 |
142 |
0 |
0 |
1 |
0 |
0 |
| Redis Scaling Horror Stories Episode 4: Night of the Living Replica
It was 3:12am when the replica rose from the dead … |
@dragonflydbio |
Company |
Original |
2026-10-26 |
134 |
0 |
0 |
0 |
0 |
0 |
| Redis Scaling Horror Stories Episode 3: The Hot Key Haunting
Our cluster had plenty of shards. Only one of them was on… |
@dragonflydbio |
Company |
Original |
2026-10-23 |
153 |
1 |
0 |
1 |
0 |
0 |
| Redis Scaling Horror Stories Episode 2: One Core to Haunt Them All
You paid for 64 cores. 63 lie dormant. One is posse… |
@dragonflydbio |
Company |
Original |
2026-10-22 |
87 |
1 |
0 |
0 |
0 |
1 |
| Maxed-out threads. Noisy neighbors. Memory spikes during backups. Creeping P99 latency.
If that sounds like your in-me… |
@dragonflydbio |
Company |
Original |
2026-10-22 |
102 |
2 |
0 |
0 |
0 |
0 |
| Struggling to scale to meet demand from AI? Learn from Hugging Face and Instacart, two hypergrowth companies that did i… |
@dragonflydbio |
Company |
Original |
2026-10-22 |
97 |
1 |
0 |
0 |
0 |
0 |
| Dailymotion is saving tens of thousands per month with Dragonfly, plus native backup, better latency and an upgrade pat… |
@dragonflydbio |
Company |
Original |
2026-10-22 |
61 |
0 |
0 |
0 |
0 |
0 |
| Redis Scaling Horror Stories Episode 1 : The Forking
Our Redis primary died in the middle of the night, and nothing in… |
@dragonflydbio |
Company |
Original |
2026-10-21 |
119 |
2 |
0 |
1 |
0 |
0 |
| No more noisy neighbors. Since moving to Dragonfly, ShareChat’s Arya Ketan says they no longer run into multi-tenancy p… |
@dragonflydbio |
Company |
Original |
2026-10-21 |
91 |
2 |
0 |
0 |
0 |
0 |
| Migrations go wrong when someone finds a problem after the point of no return. Meesho never had one: Redis to Dragonfly… |
@dragonflydbio |
Company |
Original |
2026-10-20 |
153 |
2 |
0 |
1 |
0 |
0 |
| Something’s lurking in your in-memory data store. 🦈
Spooky scaling stories from Hugging Face and Instacart, including … |
@dragonflydbio |
Company |
Original |
2026-10-19 |
239 |
6 |
0 |
0 |
0 |
1 |
| When AI growth breaks your data infrastructure, the in-memory layer often goes first. See how Instacart and Hugging Fac… |
@dragonflydbio |
Company |
Original |
2026-10-16 |
164 |
1 |
0 |
0 |
0 |
0 |
| Agent memory is the data AI agents use to retain, recall, and use information over time. One entertainment company want… |
@dragonflydbio |
Company |
Original |
2026-10-15 |
99 |
0 |
0 |
0 |
0 |
0 |
| Fewer machines. Better utilized. Meesho’s cost is down around 60% after moving to Dragonfly, with latency down 40 to 50… |
@dragonflydbio |
Company |
Original |
2026-10-14 |
165 |
1 |
0 |
0 |
0 |
0 |
| AI agents make us multi-threaded. Instead of visiting an app once a week for ten minutes, they check in every hour for … |
@dragonflydbio |
Company |
Original |
2026-10-13 |
218 |
1 |
0 |
0 |
0 |
0 |
| How do you know a new database is ready for millions of hotel searches? Agoda tested it over a real network first. Arou… |
@dragonflydbio |
Company |
Original |
2026-10-08 |
109 |
0 |
0 |
0 |
0 |
0 |
| Agoda cut their slowest database reads by 10x and now run the same workload on about half the servers. Clarkson Chang, … |
@dragonflydbio |
Company |
Original |
2026-10-07 |
154 |
1 |
0 |
0 |
0 |
0 |
| Instacart, Hugging Face, and xAI all run on Dragonfly. Instacart's results: 70% fewer nodes and 50% lower latency, serv… |
@dragonflydbio |
Company |
Original |
2026-10-07 |
190 |
2 |
0 |
0 |
0 |
1 |
| ✈️ 🇮🇪 We are headed to Dublin for SRECon EMEA, Oct 13–15, with Dragonfly 2.0 in tow: 54% more throughput, 30–40% less m… |
@dragonflydbio |
Company |
Original |
2026-10-06 |
260 |
2 |
0 |
0 |
0 |
0 |
| Dragonfly 2.0 is out today. Against 1.0 it pushes 54% more throughput, cuts average latency 35%, and uses 30 to 40% les… |
@dragonflydbio |
Company |
Original |
2026-09-17 |
448 |
6 |
0 |
0 |
0 |
1 |
| Dragonfly Ascent is almost here. Hear how @Meesho_Official, @agoda, @Dailymotion and @sharechatapp run Dragonfly in pro… |
@dragonflydbio |
Company |
Original |
2026-09-15 |
157 |
1 |
1 |
0 |
0 |
0 |
| Two keynotes, one day at Dragonfly Ascent. Roman Gershman on three years of hardening Dragonfly since 1.0. Oded Poncz o… |
@dragonflydbio |
Company |
Original |
2026-09-14 |
215 |
2 |
0 |
0 |
0 |
0 |
| Arya Ketan (@sharechatapp) on modernizing infrastructure and cutting costs ahead of an IPO, without slowing down. Drago… |
@dragonflydbio |
Company |
Original |
2026-09-10 |
246 |
3 |
2 |
0 |
0 |
0 |
| RT @romanger: "You can have a second computer once you have shown you know how to use the first one."
This quote perfe… |
@dragonflydbio |
Company |
Repost |
2026-09-03 |
12 |
0 |
3 |
0 |
0 |
0 |
| RT @OPoncz: AI is breaking cloud infrastructure.
We built systems for human scale, but AI agents generate 100x more t… |
@dragonflydbio |
Company |
Repost |
2026-09-03 |
11 |
0 |
2 |
0 |
0 |
0 |
| Session preview: @agoda on how they scaled their price cache with DragonflyDB. Live at Dragonfly Ascent, Sept 16. Regis… |
@dragonflydbio |
Company |
Original |
2026-08-31 |
371 |
5 |
1 |
0 |
0 |
0 |
| Session preview: @Dailymotion on how they consolidated Memcached, Redis, and Aerospike key-value datastores onto Dragon… |
@dragonflydbio |
Company |
Original |
2026-08-27 |
181 |
3 |
1 |
0 |
0 |
0 |
| Session preview: “Powering E-commerce at Scale, Meesho’s Dragonfly Journey.” Meesho joins us at Dragonfly Ascent to sha… |
@dragonflydbio |
Company |
Original |
2026-08-25 |
188 |
2 |
0 |
1 |
0 |
0 |
| Speaker spotlight: Arya Ketan, Distinguished Engineer at @sharechatapp, joins us at Dragonfly Ascent on Sept 16. Regist… |
@dragonflydbio |
Company |
Original |
2026-08-20 |
133 |
3 |
2 |
0 |
0 |
0 |
| Speaker spotlight: Clarkson Chang, Lead Software Engineer at Agoda, joins us at Dragonfly Ascent on Sept 16. Register: … |
@dragonflydbio |
Company |
Original |
2026-08-18 |
153 |
2 |
1 |
0 |
0 |
0 |
| Speaker spotlight: Cyril Corbon, Principal DevOps Architect at @dailymotion, joins us at Dragonfly Ascent on Sept 16. R… |
@dragonflydbio |
Company |
Original |
2026-08-14 |
189 |
2 |
0 |
0 |
0 |
0 |
| Redis doubles its hash table in memory before moving a single key.
Dragonfly grows one small piece at a time instead. … |
@dragonflydbio |
Company |
Original |
2026-08-14 |
255 |
7 |
1 |
0 |
0 |
0 |
| Dragonfly SSD Data Tiering is now available in Dragonfly Cloud.
Run a working set far larger than you could before with… |
@dragonflydbio |
Company |
Original |
2026-08-12 |
430 |
6 |
1 |
0 |
1 |
0 |
| The migration pattern for TB-scale systems with a strict SLA isn't the simple one. It's Dual Write.
Write to both the … |
@dragonflydbio |
Company |
Original |
2026-08-08 |
190 |
3 |
1 |
0 |
0 |
1 |
| Second angle on memory-aware rebalancing, this time the mechanics: Dragonfly Cloud tracks memory at the individual hash… |
@dragonflydbio |
Company |
Original |
2026-08-07 |
198 |
4 |
1 |
1 |
0 |
1 |
| Speaker spotlight: Yash Shah, Software Development Engineer at @Meesho_Official, joins us at Dragonfly Ascent on Sept 1… |
@dragonflydbio |
Company |
Original |
2026-08-06 |
142 |
4 |
2 |
0 |
0 |
0 |
| Old Dragonfly replication design: one big Set or Hash held a single lock for its entire flush. Even while waiting on th… |
@dragonflydbio |
Company |
Original |
2026-08-06 |
284 |
5 |
2 |
0 |
0 |
0 |
| A typical LlamaIndex app needs four separate storage backends: one for chat history, one for the document store, one fo… |
@dragonflydbio |
Company |
Original |
2026-08-04 |
227 |
4 |
1 |
0 |
0 |
0 |
| Three years ago we made a bet.
Our co-founder Roman had been a principal engineer on the ElastiCache team at AWS, and … |
@dragonflydbio |
Company |
Original |
2026-08-03 |
474 |
11 |
3 |
0 |
0 |
0 |
| ShareChat, a major social media platform in India, was fighting an unreliable in-memory cache: flickering availability,… |
@dragonflydbio |
Company |
Original |
2026-08-01 |
167 |
2 |
0 |
0 |
0 |
0 |
| Real-time betting odds, hundreds of thousands of concurrent users, and a 200ms delay means someone missed the number th… |
@dragonflydbio |
Company |
Original |
2026-07-31 |
442 |
8 |
0 |
0 |
0 |
0 |
| Code Red: Migration Under Pressure is live on YouTube.
Pressure is a privilege: once your caching system costs more th… |
@dragonflydbio |
Company |
Original |
2026-07-28 |
209 |
5 |
0 |
0 |
0 |
0 |
| Instacart's managed Valkey solution forced its ad-serving feature store into a multi-hundred-node deployment split acro… |
@dragonflydbio |
Company |
Original |
2026-07-24 |
190 |
4 |
0 |
0 |
0 |
0 |
| Celery and Sidekiq queues are lists of small JSON records with a fixed schema: same keys, different values, over and ov… |
@dragonflydbio |
Company |
Original |
2026-07-24 |
380 |
4 |
0 |
0 |
0 |
0 |
| AWS ElastiCache's cache.t4g.small advertises "up to 5 Gbps." What that hides: baseline guaranteed bandwidth is tiny, an… |
@dragonflydbio |
Company |
Original |
2026-07-23 |
133 |
2 |
1 |
0 |
0 |
0 |
| Redis was built single-threaded. Modern cloud servers ship 24 to 96 cores. Redis uses one of them.
At multi-terabyte s… |
@dragonflydbio |
Company |
Original |
2026-07-21 |
138 |
8 |
0 |
0 |
0 |
1 |
| An enterprise GitOps platform's heaviest Argo CD workload: 47 sources, some cache keys around 20MB, hammering Redis HA … |
@dragonflydbio |
Company |
Original |
2026-07-20 |
142 |
5 |
2 |
0 |
0 |
2 |
| Live today: Code Red: Migration Under Pressure.
Built for teams whose Redis migration just became this quarter's probl… |
@dragonflydbio |
Company |
Original |
2026-07-17 |
149 |
3 |
0 |
0 |
0 |
0 |
| +258% ops/sec vs Redis. +95% vs Valkey 9.0. Same 8-core box, same Hash workload.
Redbus (one of the world's largest on… |
@dragonflydbio |
Company |
Original |
2026-07-16 |
190 |
6 |
1 |
0 |
0 |
1 |
| Tens of millions of instruments. Exact lookup, time-series pricing, and semantic search, from one store. Here's what th… |
@dragonflydbio |
Company |
Original |
2026-07-15 |
121 |
3 |
0 |
0 |
0 |
0 |
| Valkey 9.0 fixed the I/O path. It didn't fix hashtable rehash. Writes still stall while the table grows.
Fresh benchma… |
@dragonflydbio |
Company |
Original |
2026-07-14 |
255 |
4 |
0 |
0 |
0 |
1 |
| Dragonfly tiered storage benchmark:
Full in-memory → 3.5 GB RSS, 150K keys/sec
1 GB cap + offload → 100 MB RSS, 125K k… |
@dragonflydbio |
Company |
Original |
2026-07-14 |
135 |
4 |
1 |
0 |
0 |
0 |
| Teams migrate caching layers for two very different reasons: because they wanted to, or because a licensing change or a… |
@dragonflydbio |
Company |
Original |
2026-07-10 |
146 |
5 |
1 |
0 |
0 |
0 |
| Most Redis to Dragonfly migrations do not need a maintenance window, and most teams over-plan for downtime that will no… |
@dragonflydbio |
Company |
Original |
2026-07-09 |
181 |
6 |
1 |
0 |
0 |
1 |
| 5 migration patterns from Redis to Dragonfly. Which one fits your system?
Pattern 1: Cold-Start Cutover.
Update the en… |
@dragonflydbio |
Company |
Original |
2026-07-08 |
170 |
3 |
1 |
0 |
0 |
1 |
| Valkey 9.0's I/O threading is the real deal on reads. At 8 vCPU, read-heavy workloads: within 5% of Dragonfly. That's a… |
@dragonflydbio |
Company |
Original |
2026-07-07 |
308 |
5 |
1 |
0 |
0 |
1 |
| Redbus's snapshot files are 84% smaller on Dragonfly than Redis. They write 94% faster.
That's not a code optimization… |
@dragonflydbio |
Company |
Original |
2026-07-06 |
271 |
4 |
2 |
0 |
0 |
2 |
| 6 Redis pods down to 2. 43% fewer pods. 33% fewer containers. 14% lower CPU. 20% lower memory.
This is what happened w… |
@dragonflydbio |
Company |
Original |
2026-07-03 |
345 |
3 |
0 |
0 |
0 |
1 |
| 10 isolated workloads on one Dragonfly instance. 4 lines of ACL config. No extra pods.
The standard approach for multi… |
@dragonflydbio |
Company |
Original |
2026-07-02 |
307 |
2 |
1 |
0 |
0 |
1 |
| Dragonfly v1.37 changed from per-shard HNSW to a single global HNSW index. The benchmark headline was 7x throughput, 65… |
@dragonflydbio |
Company |
Original |
2026-07-01 |
505 |
9 |
0 |
2 |
0 |
1 |
| 7.3x more sorted set throughput on the same 8-core machine. Dragonfly vs Valkey 9.0.
The setup: two AWS c8g.2xlarge in… |
@dragonflydbio |
Company |
Original |
2026-06-30 |
347 |
4 |
0 |
1 |
0 |
1 |
| Your cache doesn't fit in RAM anymore. Redis says reshard. Here's what that means: more nodes, more Sentinel processes,… |
@dragonflydbio |
Company |
Original |
2026-06-29 |
418 |
5 |
0 |
0 |
1 |
1 |
| Valkey is a fork. Forks inherit architectural debt. Here's what that means specifically for cluster rebalancing.
Redis… |
@dragonflydbio |
Company |
Original |
2026-06-26 |
282 |
5 |
1 |
1 |
0 |
2 |
| Redis Sentinel detects a hardware failure in 30-60 seconds. Dragonfly Cloud detects it in 10. Here's the dual-path mech… |
@dragonflydbio |
Company |
Original |
2026-06-25 |
223 |
4 |
2 |
1 |
0 |
2 |
| 239% more ops/sec. 70% lower avg latency. 29% less RAM. This is what Redbus found when they benchmarked Dragonfly again… |
@dragonflydbio |
Company |
Original |
2026-06-22 |
237 |
5 |
0 |
0 |
0 |
1 |
| Valkey 9.0's I/O threading is real engineering. But command execution still runs on one core.
Past ~8 io-threads, more… |
@dragonflydbio |
Company |
Original |
2026-06-18 |
258 |
3 |
2 |
1 |
0 |
1 |
| Dragonfly 1.39 is here with major new search capabilities, performance improvements (as always), and a lot more.
Searc… |
@dragonflydbio |
Company |
Original |
2026-06-11 |
130 |
2 |
0 |
0 |
0 |
0 |
| Dragonfly has been named to the 2026 @Redpoint InfraRed 100, the 100 private companies defining the next era of cloud i… |
@dragonflydbio |
Company |
Original |
2026-05-27 |
210 |
3 |
2 |
0 |
0 |
1 |
| Redis was built in a single-threaded era. Modern cloud servers have 64, 96+ cores. Redis uses one of them.
At multi-te… |
@dragonflydbio |
Company |
Original |
2026-03-24 |
603 |
2 |
2 |
1 |
0 |
2 |
| 🎉 30,000 stars on GitHub! ⭐
From launch in June 2022 → 10k stars in just 75 days → 30k stars.
Massive thanks to our g… |
@dragonflydbio |
Company |
Original |
2026-02-19 |
352 |
4 |
2 |
0 |
0 |
1 |
| Learn how @akuityio boosted Argo CD performance and simplified infrastructure by moving from Redis to Dragonfly.
A gre… |
@dragonflydbio |
Company |
Original |
2026-02-04 |
218 |
4 |
0 |
0 |
0 |
1 |
| Day 2 at #AWSreInvent is in full swing. 🚀 Ready for you at Booth 1855! Come catch Dragonfly for exclusive demos, great … |
@dragonflydbio |
Company |
Original |
2025-12-02 |
237 |
2 |
0 |
0 |
0 |
0 |
| We’re breaking the memory barrier! 🚀 Introducing data tiering in Dragonfly, extending RAM with SSDs to handle massive d… |
@dragonflydbio |
Company |
Original |
2025-11-25 |
376 |
5 |
2 |
0 |
0 |
1 |
| From Memcached to Redis/Valkey, what’s next for in-memory data? Demands of AI-powered workloads are creating a new scal… |
@dragonflydbio |
Company |
Original |
2025-11-17 |
568 |
1 |
0 |
0 |
0 |
0 |
| Context building, data replay & validation, multi-agent coordination... Sounds like enterprise #AI problems? Yes, b… |
@dragonflydbio |
Company |
Original |
2025-11-10 |
732 |
3 |
2 |
1 |
0 |
0 |
| The in-memory data store ecosystem is moving faster than ever. Our new analysis explores the landscape 1.5 years after … |
@dragonflydbio |
Company |
Original |
2025-11-06 |
375 |
3 |
0 |
0 |
1 |
1 |
| New from #MDISummit: This week’s standout talk reveals how @ScyllaDB’s tablet-based data distribution and raft-based me… |
@dragonflydbio |
Company |
Original |
2025-11-03 |
1,000 |
2 |
1 |
1 |
0 |
0 |
