azurezyq

11 ms·

31.
by azurezyq 4y ago
Add one: Got to know that cross AZ traffic is double charged (OUT and IN), making it twice as expensive as GCP.
32.
by azurezyq 4y ago
This is actually a great observation. Data pipelines are often written in various languages, running on heterogenous systems, with different time alignment schemes. I always found it tricky to "fully trust" a piece of result. Hmm,
33.
by azurezyq 4y ago
https://github.com/facebookincubator/AITemplate
34.
AITemplate, a revolutionary new inference engine by Meta AI(ai.facebook.com)
73 points by azurezyq 4y ago|35 comments
35.
by azurezyq 4y ago
1 operator per cluster is not ideal since most clusters are "stable" and don't need much care. Having plenty of them should be a headache. The operator crash on our side does sound scary. But as a DBaaS system, as long as the
36.
by azurezyq 4y ago
It actually works better for us. The system is definitely complex, but we still have some ways to debug and develop locally. For example: * You can test different parts of the system individually, via APIs. * With k8s operator model, it&#x2
37.
by azurezyq 4y ago
We are currently building a Database-as-a-service platform (tidbcloud.com) using Kubernetes. I have to say it is a love-and-hate story. On the bright side, k8s is almost the only option of an abstraction layer on top of different Clouds, fo
38.
by azurezyq 4y ago
The increasing use of bots may be due to an increased use of Infra-as-codes or other integrations. Git is no longer a mere code management system. It's already part of the production infrastructure. In the current company, we have gith
39.
Speeding Up a SaaS Platform by 5x in a Day Without Changing the Code(en.pingcap.com)
9 points by azurezyq 4y ago|4 comments
40.
by azurezyq 4y ago
I can share mine. It's an ads retrieval system. Latency is very sensitive and it has to be efficient. To avoid mem allocations, special hashtables with fixed number of buckets (also open addressing) are used in multiple places in query
41.
by azurezyq 4y ago
I have a very similar experience, but with a different profiling tool. When I first graduated from school and joined a big internet company, I'm not that "different". The serving stack was all in C++. My colleagues were reall
42.
Price to Performance Comparison on Tau T2A and AWS Graviton2 ARM-Based Instances(en.pingcap.com)
6 points by azurezyq 4y ago|0 comments
43.
by azurezyq 4y ago
Then there'll be a forked Android ecosystem from AOSP.
44.
by azurezyq 4y ago
according to https://en.pingcap.com/case-study/lesson-learned-from-querie... , it can store 1.3 trillion rows in a single DB. And in one use case: "in this application, data increases by 8 billion records and 1.5 T
45.
by azurezyq 4y ago
Yes, hardware has limits and you just cannot have a single machine with unlimited number of cores and disks. But as the article mentioned, NewSQL is already there. DBs in this category has actually been awhile. I think Spanner's paper
46.
by azurezyq 4y ago
If DB works beautifully, it should not be a problem. But in reality it doesn't. Also DB doesn't live in vacuum, so you are facing these challenges: * ORM bypass, for better performance. * Sharding and application level handling of
47.
by azurezyq 4y ago
Databases are "heavy", something once chosen, seldom changes. Many companies just stick to whatever DB stack they chose very early into the development. Many tech blogs say you can use NoSQL to solve A, relax your consistency requ
48.
by azurezyq 4y ago
If you are looking for a production ready distributed store written in Rust. Check out TiKV( https://github.com/tikv/tikv ), which was also mentioned in the acknowledge section of the project's README. There's
49.
by azurezyq 4y ago
There are "clouds" with lower networking costs, like Digitalocean/linode. But they are not "capable" enough to support a sizable company. Actually, networking is like a "tax" on top of the whole platform.
50.
by azurezyq 4y ago
I think you are talking about HTAP( https://en.wikipedia.org/wiki/Hybrid_transactional/analytica... ). There are already decent products on the market. But don't expect 100% analytical capabilities like BigQuer
51.
by azurezyq 4y ago
It's not true. For example: https://en.m.wikipedia.org/wiki/Haiyang_Nuclear_Power_Plant
52.
by azurezyq 4y ago
Raft is the secret sauce behind many distributed databases, like CockroachDB, TiDB, Yugabyte, etc.. But to make it scalable and production grade, you have to run multiple Raft operations in parallel and manage them well. E.g., sharding th
53.
by azurezyq 4y ago
Most of them are generated on the fly, like protos. But you know, there are tons of configs, all kinds of.
54.
by azurezyq 5y ago
I think the tweet is a bit too much. I worked for GCP for several years before. Impact in promo is defined within the same group and aligned with the goals of the group. If improving the stability of k8s is among the goals, then no problem
55.
by azurezyq 5y ago
Well nowadays a lot of well known websites use clickbaits regularly, e.g., wsj and NYTimes. Many times, they are willing to summarize the news in the title when the news itself is not that complicated.
56.
by azurezyq 5y ago
It seems that you don't play Genshin impact game. You can't buy ascension materials with money. You pay for stronger characters with more versatile playstyles, but it won't make the game less grindy. You still have to fight f
57.
by azurezyq 5y ago
also it seems that ART+COM already lost the lawsuit. Also I'm wondering what's the "algorithm" involved here. Isn't satellite imaginary processing and 3D rendering already matured techs by then?
58.
by azurezyq 5y ago
It doesn't look like a documentary though. Show is show.
59.
by azurezyq 5y ago
It seems that you don't use Android. There's an account system apps can use in the os. Just checked mine, I can see Google, Facebook, Reddit and quite a lot others there
60.
by azurezyq 5y ago
I used to work in GCP. The billing report UI in the billing account section shows per SKU usage. While detailed breakdown can be a nontrivial monster. Some customer may just launch thousands of VMs or data processing jobs per day.

More ›