ClickHouse and Postgres are great precisely because they refuse to compromise on opposite concepts. Some kind of hybrid HTAP database has been promised for many years and no attempt has hit the bar, even for yesterday's workloads. AI has massively raised that bar.
We won't make that compromise. We're investing in ClickHouse + Postgres, as distinct best-of-breed engines, with a unified experience that feels like one system to the user.
https://t.co/Lb4BuoNLtd
A few weeks into beta, and Postgres managed by ClickHouse has already shipped RBAC, Terraform support, ClickPipes migrations, billing, and a revamped console.
https://t.co/8YXMzGoQnl
Moving fast on making it easier to operate and more integrated with ClickHouse Cloud.
No information on who the blue and yellow vendors are. No visibility into hardware config. No configuration of the HammerDB setup - warehouse count, concurrency etc. Unclear if any tuning was done any of the databases. Almost impossible to reproduce these results.
Again don't mean to shame anyone in anyway. But I wish vendors were more transparent when sharing such benchmarks to of thousands of people. Keeping them opaque defeats the purpose of these benchmarks. I'm sure people will see through all of this!
Restarts and connection dropouts shouldn't be part of running Postgres at scale.
https://t.co/HKGHPQWJ5N
That's what Socialpruf ran into with Neon. After migrating to ClickHouse Managed Postgres, the outages stopped, and performance improved.
. Our Postgres service at @ClickHouseDB has fully managed migration tooling and this thing works at TB scale. Give it a shot and try to break it! 🤗
Great work @AmoghBhara81230 for all the work on this! It has been one of the most useful and favorite features for customers. 🙏
. @asahidex What a great blog on stress-testing Postgres CDC with managed Postgres and @ClickHouseDB in ClickHouse Cloud. Great work! 👏👏
The bottleneck eventually becomes logical decoding, and even 30K–50K rows/sec per replication slot is a very high throughput rate (though it can vary depending on row width).
For most real-world use cases, this is more than sufficient, and sudden workload burts typically eventually catchup with mangeable lag. To scale further, you can use multiple pipes/replication slots.
And maybe we’re working on something new that could push the limits even further. 👀👀 https://t.co/yYQn3H4oo2
I'm a student from BITS Pilani doing my bachelor's project at the Univ of Auckland for the last 4mo. I have completed my work, and as planned should have returned by now. It is getting very difficult to manage my finances now. Please arrange for my return ASAP.
@IndiainNZ @MEAIn
What are they even thinking? Rising Student loans are gonna help us in anyway? The enormous sums will fetch them better rankings? Fee hike will make BITS equal to MIT?
#BITSPilaniAgainstFeeHike#RollBackBITSPilaniFeeHike