franckpachot

8 ms·

31.
by franckpachot 3y ago
(YugabyteDB Developer Advocate here) @flagged24 If you can send me more info about your migration problems, I would love to look at it (fpachot@yugabyte.com). The postgres-compatibility, performance, and YB Voyager are improving from feedba
32.
by franckpachot 3y ago
You have a misconception about CockroachDB and YugabyteDB architecture. It is not 2-way replication. It writes to a Raft group. The transaction can be picked up by any node because the transaction intents are also replicated. The main diffe
33.
by franckpachot 3y ago
Referring to Distributed SQL as a key-value store is like defining monolithic databases as a block storage. It reduces it to an internal structure. What makes it a database is what is on top: ACID transaction, full SQL features, relational
34.
by franckpachot 3y ago
Sharding above is not transparent (the application has to do it) Sharding below is what Distributed SQL does (what they call distributed key-value storage, ignoring that the SQL transactions are also distributed)
35.
by franckpachot 3y ago
Aurora writes each block to 6 replicas, and waits for a quorum of 4 to acknowledge. This has small impact on commit time which you can see as IO:XactSync ( https://dev.to/aws-heroes/aws-aurora-io-xactsync-is-not-a-po...
36.
by franckpachot 3y ago
All benchmarks are biased because nobody does the same level of config tuning on different databases. But this GigaOm/Microsoft one is a joke. They run Citus with no replication on 12x16+1x32 vCores and the others on 14x16 vCPU (so 14x
37.
by franckpachot 3y ago
but large cache is expensive in the cloud and you cannot scale up/down without downtime
38.
by franckpachot 3y ago
I mentioned Neon because it is open source and the article says "Not OSS" for "DBMS-optimized cloud storage". Think of it as open-source Aurora https://neon.tech/
39.
by franckpachot 3y ago
In the case of YugabyteDB, here is how we avoid "excessive number" of networks hops - true Index Only Scan. PostgreSQL doesn't store the MVCC visibility in indexes and have to look at the table even in case of Index Only Scan
40.
by franckpachot 3y ago
The article is a marketing promotion for Citus. They ignore Neon, a major innovation in distributed cloud storage solutions. They reduce Distributed SQL to key-value stores, when it is actually the only distributed architecture that preserv
41.
by franckpachot 3y ago
The "disappointing" benchmark mentioned in the article is a shame for GigaOm who published it and for Microsoft who paid for it. They compare Citus with no HA to CockroachDB and YugabyteDB with replication factor 3 Multi-AZ, resil
42.
by franckpachot 3y ago
It is in the Wiki: https://wiki.postgresql.org/wiki/Serializable Any transaction which is run at a transaction isolation level other than SERIALIZABLE will not be affected by SSI. If you want to enforce business rules
43.
by franckpachot 3y ago
What is monolithic in traditional SQL databases when compared to Distributed SQL?
44.
by franckpachot 3y ago
Interesting, that's the one I don't really agree with. Bottleneck, yes, because it is a monolithic stream for the whole database. But corruption?
45.
by franckpachot 3y ago
Yes, everything Yugabyte does is open source
46.
Improving PostgreSQL: How to Overcome the Tough Challenges(yugabyte.com)
29 points by franckpachot 3y ago|6 comments
47.
by franckpachot 3y ago
This is incredibly awesome. When Oracle started to talk about Javascript in the database (around 2016), with GraalVM and MLE, it was clear for me that the goal was not to deploy stored procedures in Javascript but, at some point have the JS
48.
by franckpachot 3y ago
Yes, CockroachDB and YugabyteDB do more than sharding, they are distributed SQL (difference here: https://dev.to/yugabyte/partitioning-vs-sharding-what-about-... ). YugabyteDB is open source and has more postgres featur
49.
by franckpachot 3y ago
Looking at the execution plan, IN() or =ANY() seems faster: https://dbfiddle.uk/zIk9IrEN OR reads the same number of buffers, but with a long execution plan
50.
by franckpachot 3y ago
The power of SQL is that it is structured, based on mathematic definitions. Once written and tested, the surface of possible bug or wrong result is very limited. LLMs are based on language interpretation and do not have this precision
51.
by franckpachot 3y ago
I'm not sure those are great recommendations. First, the range scan on Tid is there only since PostgreSQL 14: https://commitfest.postgresql.org/32/2924/ And the CTID changes on an active database: https:&#x2
52.
by franckpachot 3y ago
With an index on User (createdAt, id) and one on ContactMethod ( primary,ContactMethod,userId), it should be fast (check that the the execution plan starts with User). Except if lot of recent users have no phones, but that will not be bette
53.
by franckpachot 3y ago
Many decision trees I've seen when NoSQL purpose-built were becoming popular are obsolete now that the main SQL databases have structures, APIs and optimisations for non-relational models: JSONB, Text Search, Timeseries, Documents... T
54.
by franckpachot 3y ago
I used your example to show how the 2 indexes solution is better on PostgreSQL, but the one composite index is the best one for YugabyteDB (PostgreSQL on a Spanner-like architecture): https://dev.to/yugabyte/one-fat-ind
55.
by franckpachot 3y ago
by itself, Citus is just sharding to multiple databases. Then each of this database can be protected like any database with a standby
56.
by franckpachot 3y ago
Bitmap Scan relies on shared buffers. Distributed SQL cannot share the buffers between nodes. BitmapAnd is nice but expensive. I prefer a compound index on foo(bar, baz) but PostgreSQL needs an additional index if there are query with condi
57.
by franckpachot 3y ago
Citus is not close to Spanner. No global secondary indexes, no cross-shard ACID (the commit status is eventually consistent). No auto-resharding. Yugabyte and Cockroach have a spanner-like architecture, but different open-source model and p
58.
by franckpachot 3y ago
23c Free is the same as the previous Oracle XE. Free (really, no risk, can be deployed anywhere) but limited in resources. The risk is to develop your application with all features and then have to buy all options for production
59.
by franckpachot 3y ago
On self-hosted you can use the same as what cloud vendors are doing. Patroni or pg_auto_failover to manage single-primary + replicas. Maybe Neon to run an Aurora-like. CitusData... Azure did a lot around to get elasticity. Not easy to do th
60.
by franckpachot 3y ago
Google has Spanner. AWS is working on something similar. And both have YugabyteDB in their marketplace. Those are Distributed SQL (Global ACID), not Citus. For DataWarehouse which doesn't need ACID, there are other services.

More ›