Mastodon
-
@JaredBusch said in Mastodon:
So I jumped around a few random tags to see what was out there.
ended up on the #monday tag https://mastodon.social/tags/monday
I saw a post there from 2 months ago form a "news feed"
Clicked on the name to jump to their feed and see what it said.
That took me to here https://quitter.im/newsstream
and like many of the other random names I clicked on they all reference the GNUSocial network of federated servers and the Ostatus protocol.
-
From that page, I clicked on the gnusocial header icon
That took me here
https://quitter.im/main/allclicking on the posts there expands them to show they are coming from many different servers.
-
When I log in, it shows a federated timeline:
-
Gargron (Mastodon author and owner of mastodon.social) posted on Medium about his experiences scaling.
-
@aidan_walsh it's all running on bare metal!
-
-
@JaredBusch said in Mastodon:
@aidan_walsh it's all running on bare metal!
I get that the guy is a software developer so he's not IT it's not his job to understand how bad that is but as we are IT, someone so much need to go help fix the design.
-
@JaredBusch said in Mastodon:
@JaredBusch said in Mastodon:
@aidan_walsh it's all running on bare metal!
I get that the guy is a software developer so he's not IT it's not his job to understand how bad that is but as we are IT, someone so much need to go help fix the design.
Wasn't this the issue that Github had too? They have a developer in an infrastructure position and then lost a bunch of data? Seems like this would be a similar situation.
-
@JaredBusch said in Mastodon:
@JaredBusch said in Mastodon:
@aidan_walsh it's all running on bare metal!
I get that the guy is a software developer so he's not IT it's not his job to understand how bad that is but as we are IT, someone so much need to go help fix the design.
Any I feel like PostgreSQL is an odd choice for this. Relational data for micro-blogging? He talks about issues scaling the database... issues that appear to be caused specifically by choosing a relational database like this.
-
@JaredBusch said in Mastodon:
@JaredBusch said in Mastodon:
@aidan_walsh it's all running on bare metal!
I get that the guy is a software developer so he's not IT it's not his job to understand how bad that is but as we are IT, someone so much need to go help fix the design.
Wasn't this the issue that Github had too? They have a developer in an infrastructure position and then lost a bunch of data? Seems like this would be a similar situation.
Yes, sounds very familiar.
-
@scottalanmiller said in Mastodon:
@JaredBusch said in Mastodon:
@JaredBusch said in Mastodon:
@aidan_walsh it's all running on bare metal!
I get that the guy is a software developer so he's not IT it's not his job to understand how bad that is but as we are IT, someone so much need to go help fix the design.
Wasn't this the issue that Github had too? They have a developer in an infrastructure position and then lost a bunch of data? Seems like this would be a similar situation.
Yes, sounds very familiar.
Except this is not at least to my understanding a large company pretty much one guy spinning up some stuff trying to start something and he doesn't know any better because he's not IT is simply the guy that develop the software
-
@JaredBusch said in Mastodon:
@scottalanmiller said in Mastodon:
@JaredBusch said in Mastodon:
@JaredBusch said in Mastodon:
@aidan_walsh it's all running on bare metal!
I get that the guy is a software developer so he's not IT it's not his job to understand how bad that is but as we are IT, someone so much need to go help fix the design.
Wasn't this the issue that Github had too? They have a developer in an infrastructure position and then lost a bunch of data? Seems like this would be a similar situation.
Yes, sounds very familiar.
Except this is not at least to my understanding a large company pretty much one guy spinning up some stuff trying to start something and he doesn't know any better because he's not IT is simply the guy that develop the software
True, that does make it a bit better.
-
Databases scale vertically. That means, it’s a lot easier and more cost efficient to buy a super beefy machine for your database, than it is to spread the database over multiple machines with sharding or replication.
Why Postgres? That seems like the least logical choice for this.
-
@scottalanmiller said in Mastodon:
@JaredBusch said in Mastodon:
@scottalanmiller said in Mastodon:
@JaredBusch said in Mastodon:
@JaredBusch said in Mastodon:
@aidan_walsh it's all running on bare metal!
I get that the guy is a software developer so he's not IT it's not his job to understand how bad that is but as we are IT, someone so much need to go help fix the design.
Wasn't this the issue that Github had too? They have a developer in an infrastructure position and then lost a bunch of data? Seems like this would be a similar situation.
Yes, sounds very familiar.
Except this is not at least to my understanding a large company pretty much one guy spinning up some stuff trying to start something and he doesn't know any better because he's not IT is simply the guy that develop the software
True, that does make it a bit better.
I am sure that mastodon that social as a company and there's more than one person but it's probably just really tiny and all for development
-
@scottalanmiller said in Mastodon:
@JaredBusch said in Mastodon:
@JaredBusch said in Mastodon:
@aidan_walsh it's all running on bare metal!
I get that the guy is a software developer so he's not IT it's not his job to understand how bad that is but as we are IT, someone so much need to go help fix the design.
Any I feel like PostgreSQL is an odd choice for this. Relational data for micro-blogging? He talks about issues scaling the database... issues that appear to be caused specifically by choosing a relational database like this.
Ha this didn't show up while I was writing my post.
-
@coliver Github or Gitlab?
-
@stacksofplates said in Mastodon:
Databases scale vertically. That means, it’s a lot easier and more cost efficient to buy a super beefy machine for your database, than it is to spread the database over multiple machines with sharding or replication.
Why Postgres? That seems like the least logical choice for this.
Not the least, it's better than something expensive or even MariaDB. It has great performance. If you never needed to scale, it might easily be best. But, as he's already found, he needs to scale.
-
-
@scottalanmiller said in Mastodon:
@stacksofplates said in Mastodon:
Databases scale vertically. That means, it’s a lot easier and more cost efficient to buy a super beefy machine for your database, than it is to spread the database over multiple machines with sharding or replication.
Why Postgres? That seems like the least logical choice for this.
Not the least, it's better than something expensive or even MariaDB. It has great performance. If you never needed to scale, it might easily be best. But, as he's already found, he needs to scale.
Ya I guess I didn't mean least. It's obviously better than Oracle or Firebird. I just meant out of the plethora of options ( I was thinking Elasticsearch or Solr) this seemed like it wouldn't be considered.
-
@stacksofplates said in Mastodon:
@scottalanmiller said in Mastodon:
@stacksofplates said in Mastodon:
Databases scale vertically. That means, it’s a lot easier and more cost efficient to buy a super beefy machine for your database, than it is to spread the database over multiple machines with sharding or replication.
Why Postgres? That seems like the least logical choice for this.
Not the least, it's better than something expensive or even MariaDB. It has great performance. If you never needed to scale, it might easily be best. But, as he's already found, he needs to scale.
Ya I guess I didn't mean least. It's obviously better than Oracle or Firebird. I just meant out of the plethora of options ( I was thinking Elasticsearch or Solr) this seemed like it wouldn't be considered.
Yeah... CouchDB, MongoDB, all kinds of things that seem like they'd be better for this.