Old company of mine was handling multi-tenancy at a schema level i.e. each new tenant got their own schema with like 50 tables. It sucks but it's a legit pattern.
AWS RDS uses My SQL 8.0 by default for new databases. Also looking at there announcement it doesn't look like they are keen on supporting 5.7 for a long time.
I can't imagine a case where I'd be running a site and assuming just a single DB query per user, an absurd assumption, where I'd go... 3 users per second, yeah that's good enough. Like WTF.
Please make a video about multi-tenant database design and how to shard it. It would be very informative. I chose MySQL over PostgreSQL because it seems it can be sharded via Vitess, whereas there isn't a good, battle-tested sharding solution for PostgreSQL.
If you’re building that for a learning experience, you can use Citus a Postgres extension. If you’re prematurely optimizing, don’t shard, lots of unnecessary headache there. If you absolutely need for it a business need in this moment, and want to elaborate on business goal(s), technical goal(s), and current setup I can be more helpful. Vertical sharding, horizontal sharding, diff ways of determing how/where to shard etc
@@Patrickdaawsome well, I actually need it for production. I have a mysql database with +300GB data and eventually in some point, i have to horizontally shard it. there is no way to escape.
@@erfanium Vitess is not for you. With only 300GB of data, Vitess is overkill and will frustrate you more than you can imagine. For multi-tenant, you need to look at functional partitioning where you move specific databases/schemas to other servers. (Ex. Customer DBs 1, 2, 3 on server A, DBs 4, 5,6 on server B, etc)
@@Y-JA Speed in doing what? Note that I am a DBA, although I've worked mostly with Oracle, SQL Server, HANA, and some Postgres (mostly with PostGIS). I'm curious what modern use-case there is for MySQL over alternatives.
Read heavy operations in MySQL are faster compared to Postgres. I guess this is because it uses row level locks and query cache(which disabled now my deafult now though as per video). Also, MySQL has been around for way longer and many people started their dev journey with something like LAMP, and are well versed with MySQL.
I think you can still get memory leaks with a lower number. Just that with 10K tables, the time will be less. TBH even I haven't seen more than 500 tables in my short professional career of 3 years so far.
The situation is worse than u depict. 28 - 36 is a disaster. 1st good release in that series was 37. All this focus on innodb cluster is crap. They need a totally different concept for the storage, like Aurora.
@@phpnotasp I work with Aurora and MySQL RDS, so I know that is simply not true. The compatibility layer is an innodb fork but the the Aurora storage system is a completely different beast than GP2/3 LVM / EBS. From what I have seen from their presentation and the results I get back, the storage system is some sort of hash pool based on dynamodb nodes. It's way more efficient and on larger data sets its cheaper. A lot cheaper than other the RDS options.
I don't see why someone would use MYSQL vs Postgres. I learned the basics of MYSQL first but the knowledge is somewhat transferable to other relational DB's. I am not a DB expert by any means so I may be incorrect I just don't see the reason to make that choice technically.
Brush up your database fundamentals databases.win
Unironically, the 2018->2014 in the thumbnail made me click the video
Me too.
I was confused for a moment, too
I thought it was a clickbait thumbnail, but fortunately just a typo.
I thought it would have some deeper meaning 😂
ops , fixed it. Hopefully that doesn't lower the clickthrough rate haha
That's called "The Oracle disease"
Aren't the MariaDB devs the original MySQL devs?
correct they went their own way after Oracle got involved with MySQL
Yep, that's why I use MariaDB.
Yep, that's true.
Nahh.. I don't use dolphin when knew the elephant has array type and has better uuid
I enjoy your podcasts so much. You have no idea how much I have learnt from you. Thank you so much.
I think you have mistakenly mentioned 2014 instead of 2024 in thumbnail :D
10k tables? you need to think again
Old company of mine was handling multi-tenancy at a schema level i.e. each new tenant got their own schema with like 50 tables. It sucks but it's a legit pattern.
The Oracle Vogons announced: "Your MySQL has been scheduled for demolition. We have no mercy at all."
AWS RDS uses My SQL 8.0 by default for new databases. Also looking at there announcement it doesn't look like they are keen on supporting 5.7 for a long time.
5.7 has been dead for 4 years. That's a pretty long time already to keep supporting dead software.
I can't imagine a case where I'd be running a site and assuming just a single DB query per user, an absurd assumption, where I'd go... 3 users per second, yeah that's good enough. Like WTF.
Please make a video about multi-tenant database design and how to shard it. It would be very informative. I chose MySQL over PostgreSQL because it seems it can be sharded via Vitess, whereas there isn't a good, battle-tested sharding solution for PostgreSQL.
If you’re building that for a learning experience, you can use Citus a Postgres extension.
If you’re prematurely optimizing, don’t shard, lots of unnecessary headache there.
If you absolutely need for it a business need in this moment, and want to elaborate on business goal(s), technical goal(s), and current setup I can be more helpful.
Vertical sharding, horizontal sharding, diff ways of determing how/where to shard etc
@@Patrickdaawsome well, I actually need it for production. I have a mysql database with +300GB data and eventually in some point, i have to horizontally shard it. there is no way to escape.
@@erfanium Vitess is not for you. With only 300GB of data, Vitess is overkill and will frustrate you more than you can imagine. For multi-tenant, you need to look at functional partitioning where you move specific databases/schemas to other servers. (Ex. Customer DBs 1, 2, 3 on server A, DBs 4, 5,6 on server B, etc)
@@phpnotasp it sounds like creating a vitess at home to me.
Why would anyone use MySQL over Postgres? What is your use-case that shows it to be better?
For Vitess
@@Y-JA Speed in doing what?
Note that I am a DBA, although I've worked mostly with Oracle, SQL Server, HANA, and some Postgres (mostly with PostGIS). I'm curious what modern use-case there is for MySQL over alternatives.
@@hallkbrdz No, not speed. Vitess. The database clustering system for MySQL built by RUclips.
@@hallkbrdz Vitess is a database clustering system.
Read heavy operations in MySQL are faster compared to Postgres. I guess this is because it uses row level locks and query cache(which disabled now my deafult now though as per video).
Also, MySQL has been around for way longer and many people started their dev journey with something like LAMP, and are well versed with MySQL.
Is the case of oracle degrading it's performance intentionally?
10000 tables is not a lot? Seems like a lot to me. 😅
bruh that's nothing
u sounds like a noob
@@sad_man_no_talent how is that nothing and what's the use case for that?
It’s not a lot, but yes it’s not little, that’s why they test it with such a number. If the tests sucks at this, you know it’s bad.
I think you can still get memory leaks with a lower number. Just that with 10K tables, the time will be less.
TBH even I haven't seen more than 500 tables in my short professional career of 3 years so far.
I hate mysquel and I LOVE postgr squel
Posquel
they want you to use mysql heatwave and move to their OCI
I thought MariaDB was MySQL. That is how much I pay attention to dbs...
They are basically the same database. MySQL is too big to bother fixing when postgres already works.
Don't say that to a MariaDB employee. They are adamant that MariaDB is not MySQL. They forked MySQL and are now their own thing.
the course is really good, a lot of details
The situation is worse than u depict. 28 - 36 is a disaster. 1st good release in that series was 37. All this focus on innodb cluster is crap. They need a totally different concept for the storage, like Aurora.
In case you didn't know, Aurora is just a fork of the InnoDB engine with redo logging replaced with DRBD.
@@phpnotasp I work with Aurora and MySQL RDS, so I know that is simply not true. The compatibility layer is an innodb fork but the the Aurora storage system is a completely different beast than GP2/3 LVM / EBS. From what I have seen from their presentation and the results I get back, the storage system is some sort of hash pool based on dynamodb nodes. It's way more efficient and on larger data sets its cheaper. A lot cheaper than other the RDS options.
Watch Anton Putra's benchmark video.
I've only used MySQL,
I don't see why someone would use MYSQL vs Postgres. I learned the basics of MYSQL first but the knowledge is somewhat transferable to other relational DB's. I am not a DB expert by any means so I may be incorrect I just don't see the reason to make that choice technically.
Vacuum still sucks ahahahahaha
Informative
Bruh u didn't even finish the video
F Microsoft and their products
MariaDB + galera cluster.
MongoDB 8 released week ago
sqlite the best