Run affordable open source ClickHouse® anywhere with #AltinityCloud, Expert Support, BYOC, & real-time data lake integration to scale queries on cheap storage.
Joined October 2015
- Tweets 2,373
- Following 365
- Followers 1,712
- Likes 983
991 Photos and videos
Pinned Tweet
24 Apr 2025
⚡ Supercharge #ClickHouse on a budget!
Project Antalya = 10x performance on 10x cheaper storage... and it's #opensource
🛠️ Repo’s live: hubs.la/Q03jJq5r0
Submit PRs, contribute, or drop a ⭐ (if you’re into that sort of thing).
4
758
Jun 12
#AI in production is already breaking things. 3 #meetups this June are tackling the messiest parts. Talks on agent safety, AI #observability, agent telemetry analytics, and product insights.
SF: hubs.la/Q04lfl4v0
SEATTLE: luma.com/59wq0z0j
LA: hubs.la/Q04lfkPF0
115
Jun 11
🚨 If an #AI agent queries #ClickHouse® on behalf of a user, the access model matters.
#AltinityAntalya 26.3 adds OAuth authorization support in Antalya builds, plus more work for Iceberg/S3 and hybrid table use cases.
📝 Release notes: hubs.la/Q04l0_tw0
89
Jun 9
Josh Lee is pulling double duty at #KCDNewYork tomorrow: lightning talk at 8:25 AM on OTel vendor neutrality in practice, then a panel at 11:20 AM on #telemetry that actually matters with Carol Valencia (KrolCloud) & Diana Todea (VictoriaMetrics).
🔗 hubs.la/Q04k2Srr0
1
41
Jun 5
Most #observability problems aren't tool problems. They're instrumentation problems. 🔍
Josh Lee is breaking it down tomorrow at #SREday NY: #OTel Arrow, sampling, filtering, schema management with Weaver, & what to actually instrument.
🔖 Bookmark it: hubs.la/Q04k2CTr0
47
Jun 4
A customer's #AI agent deleted production #ClickHouse® tables the morning of this webinar. 😬
We demoed 3 tools to prevent exactly that: #MCP read-only mode, DBA Skills, & OAuth guardrails.
➕ See how #Claude builds a Grafana dashboard from a query.
hubs.la/Q04k2LLv0
1
1
108
Jun 2
What tools does a team of #ClickHouse® experts have in their AI toolkit? 🧰
Tomorrow, the Altinity team opens the attic: MCP, skills, spec-based development, code reviews, security holes, & what #AI means for real-time analytics.
Free #webinar | 9 am: hubs.ly/Q04hCHWP0
87
May 28
🛠️ AI demos are easy. Building #AI systems that work in prod is where it gets interesting.
And that's exactly what practitioners will explore at OSA Community #meetups this June:
📍 SF: hubs.la/Q04j97gz0
📍 Seattle: hubs.la/Q04j9d050
📍 LA: hubs.la/Q04j983h0
119
May 26
CVE severity scores weren't designed to answer "should I wake someone up right now?" 🛡️
They're calculated against a generic baseline with no deployment context. Mikhail Filimonov, Principal Engineer, shows how to actually triage CVEs for #ClickHouse®: hubs.ly/Q04hDnDb0
104
May 25
#ClickHouse® deployments generate thousands of alerts a day. No team can reliably act on all of them. 😮💨
But #AI changes that. Catch @dbcicero's demo at #PerconaLive Bay Area on 5/27 (11:20 AM): AI-assisted alert analysis, root cause automation more: hubs.la/Q04gZRFH0
1
82
May 22
#FIPS 140-2 & FIPS 140-3 look the same from the outside. The #encryption library underneath is a different story. 🔒
Altinity's #ClickHouse® v25.3.8 FIPS build replaces OpenSSL w/ AWS-LC (FIPS 140-3 certified, better encrypt/decrypt perf). ARM supported: hubs.la/Q04hChnV0
82
May 21
#AI workloads don't behave like traditional apps. So why are most data systems still designed like they do? 🤔
Josh Lee will open #LLMday London as a keynote on May 28, covering how to support AI workloads that demand speed, scale, & operational clarity: hubs.la/Q04gYsK60
36
May 20
"We're using #OpenTelemetry, so we have #observability." Most teams believe that until they see costs. 👀
Josh Lee is talking about what actually comes next at #SREday Munich tomorrow (2:30 PM). Expect OTel Arrow, Weaver, sampling & filtering strategies: hubs.la/Q04gYlfH0
48
May 18
"I'm yet to meet an engineer who tells me they love database migrations." 😅
In Ep 8 of #UnevenlyDistributed podcast, @ShivjiJha sat down with Josh Lee to talk about Iceberg, #AI agents, Postgres #ClickHouse®, database optimizers, & more.
Full episode: hubs.la/Q04gZ75M0
90
May 14
Everyone's running #AI experiments. Not everyone can tell you which ones are worth running in production. 😅
On May 20 (9 am PT), we're showing what stuck: MCP, agents, & schema design & query generation against #ClickHouse® on real datasets.
Register: hubs.la/Q04ffkK10
57
May 12
#AI-powered homelabs are more than “run a model locally.” 😅
Henrik Rexed (@Dynatrace) and Josh Lee talk through the architecture, workflows, observability, and cloud vs. self-hosted tradeoffs behind the setup.
Full #UnevenlyDistributed episode 👇🏻
hubs.la/Q04gcVcd0
73
May 11
#ClickHouse® migration gotcha: after rsyncing to multiple replicas, delete the server UUID file from each destination. 🛠️
Otherwise replicas sharing a #ZooKeeper ensemble will break each other's sessions & it's hard to debug.
Full rsync #migration guide: hubs.la/Q04fZ0Qm0
2
153
May 8
Quick Q: when does #ClickHouse® ReplacingMergeTree actually deduplicate? 🤔
If your answer is "when data is inserted," then come to our May Office Hours.
📅 May 20 | 8 am PT
❗ Bring your edge cases
🙋🏼 Submit Qs to #officehours on Slack/register here:
hubs.la/Q04fYVW00
69
May 7
Real-time analytics on cheap object storage, without tradeoff. 🔍
#AltinityAntalya 25.8.16.20002 adds GC support for #datalakes, icebergLocalCluster, EXPORT PARTITION for ReplicatedMergeTree, Hybrid table engine improvements, & observability updates.
hubs.ly/Q04ffBD20
1
72
May 5
The question every data team is asking: what #AI tools work for real-time analytics? 🤔
We'll demo it live on 5/20: MCP, agents, AI-generated schema design & queries running against real datasets in #ClickHouse®, & how to apply AI safely in prod.
💪🏻 hubs.la/Q04ffx910
1
1
78
May 4
You have 36 cores, 1TB of RAM, and fast NVMe disks. But your #ClickHouse® merges are still slow. Why? 🤔
The culprit is usually background_pool_size sitting at default while your hardware goes to waste.
Get the full tuning guide with config snippet: hubs.la/Q04ffD2r0
2
115