nikita

21 ms·

241.
by nikita 4y ago
Please email beta@neon.tech and we will bump you up on the waitlist. Portability is key - we can't fork Postgres. Apart from branching it's the same Postgres - so you can develop locally against vanilla Postgres. We are thinking h
242.
by nikita 4y ago
Heikki was behind major decision and has done a lot of prototyping. Happy to add more color
243.
Architecture Decisions in Neon(neon.tech)
48 points by nikita 4y ago|7 comments
244.
by nikita 4y ago
Time an again AI systems outperform humans. This maybe true today but not tomorrow
245.
by nikita 4y ago
This is awesome idea! I'm sure there is an implementation somewhere.
246.
by nikita 4y ago
Traditionally query optimizers are focused on producing good and not disastrous plans. Not necessarily the best plan. This is the role of a query optimizer: do decently well and don't royally screw up. What makes optimizers to screw up
247.
by nikita 4y ago
Disk space is becoming less of an issue in cloud environment with the rise of multi-tenant storage: aurora, neon (i'm ceo of neon). The cost of index maintenance is real and should be considered. Due to the multivariable nature of the
248.
by nikita 4y ago
One of the reasons is "safety". It's hard to make sure that adding an index won't have adverse effects on your app. Adding an index invalidates query plans and new plans may or may not be as good as the old ones. I think
249.
by nikita 4y ago
The author of the presentation is here. Happy to answer questions.
250.
by nikita 4y ago
It’s possible, but you will need dive deep into how the machinery works. The code is there
251.
by nikita 4y ago
CRDTs are pretty hard to implement at the Neon storage level - pages don’t know what’s written on them. I don’t know of a general purpose database that supports CRDTs. It would be cool though!
252.
by nikita 4y ago
Yes Aurora gets really expensive on IOPs. They basically charge you for pulling pages from their own page servers. In their design they collocate page server and safekeeper functionality. I believe we will be cheaper in this case yes.
253.
by nikita 4y ago
There is no reason for the safe keepers that run consensus SSDs to be slower than EBS which is network attached storage.
254.
by nikita 4y ago
There is a buffer cache in the compute instance. There is also cache at the page server. Postgres also uses a file system cache - that’s gone.
255.
by nikita 4y ago
Let us bump you up. We received a lot of sign ups. Mostly due to the HN power and now onboarding as fast as we can as well as fixing small issues please email at beta@neon.tech. We will be asking for feedback in return.
256.
by nikita 4y ago
Yes it’s a lot of work. But a lot less compared to building a database from scratch. There is a blog coming out from Heikki and it covers the tricky spots
257.
by nikita 4y ago
Yes you can’t infinitely scale writes. You can make your storage bandwidth infinit but you are still limited to the how much Postgres WAL one instance can pump into the storage. I think we can do a lot of good things here over time and have
258.
by nikita 4y ago
Don’t forget to star us there :)
259.
by nikita 4y ago
We can start working with you on this please email beta@neon.tech
260.
by nikita 4y ago
So you can selfhost our storage and we are making it easier and easier. Mostly we want to write one helm chart that will package it all. We are not planning to support on prem deployments commercially. We are working with partners like Perc
261.
by nikita 4y ago
This is a tricky problem to solve. Reverting to the commit - that's easy and we have all the infrastructure already to allow you to do that. The tricky part is * Make a schema change * push into prod * accumulate some new data * revert
262.
by nikita 4y ago
We will take a note of that. We are all catching our breath now. More will of course come! Thank you for pointing this out.
263.
by nikita 4y ago
We are not ready to answer this because pricing is always a model that is built on top of your COGS (cost of goods sold) and validated on real usage. We just started to onboard users - we have hundreds now, but it's just the first week
264.
by nikita 4y ago
All the features. It's the same Postgres that you use elsewhere
265.
by nikita 4y ago
I'm in contact with Yuri - CEO of EdgeDB. I'll make sure he sees this comment.
266.
by nikita 4y ago
We shut down the old compute and start a new one with a new version (after loads of testing and making sure it'll work).
267.
by nikita 4y ago
For many workloads it’s on par. Few are slower. Pathological case, working set doesn’t fit in memory allotted to pg but fits in file system page cache. There is not page cache in neon
268.
by nikita 4y ago
Please apply! Are cost is location adjusted. In most areas we pay 90th percentile
269.
by nikita 4y ago
It corresponds to Postgres 8k page. We store them in an LSM tree so we can move layers into S3
270.
by nikita 4y ago
Tips for transitioning to database development: learn rust, start working on systems, ideally get a systems job, and optimize for being prolific. Write a lot of code

More ›