MySQL Support

[roaku 795]

2 hours ago, rsvg said:

Came here to show my support for this very basic concept that Emby is purposefully avoiding (it seems so that they can monetize this in the future according to my thread that was immediately locked). 

Please consider adding this in order to retain users. It seems like a no brainer when a container image of software is available that people may want to use it in an HA (k8s) manner.

Looks like your thread was locked *after* the preexisting discussion threads like this one were shared and *after* everyone was explicitly invited to continue the discussion here after the more subtle hint was missed.

And a more fair interpretation of what you were told about the feature is that because it's a major architectural change and not necessary for the majority of their users (according to their analysis), the only way it would make sense for them to spend the significant development time required to add it would be to *create* a customer base that would need and pay for features like that and allow them to recoup those costs.

 

Edited February 24, 2021 by roaku

[byakuya32 18]

Microsoft sql would be nice too rather than a proprietary database engine it has better performance too.

[rsvg 14]

Quote

after the more subtle hint was missed.

 

didn't miss any subtle hint. I simply responded to someone else that asked a question after they showed me similar threads. 

You didn't need to reword it for them, I understood completely. While many modern software solutions use an ORM so that the user can actually choose their database (mongo, postgres, mysql, etc), Emby's architecture, on the other hand is so fundamentally bound to an embedded database, that it's developers are completely incapable of enabling the concept of an external database, even though a huge majority of the syntax is exactly the same between SQLite and MySQL. And even if it were possible, development of that feature would be monetized and sold to enterprises first. 

Thanks for clarifying?

Edited February 24, 2021 by rsvg

[roaku 795]

3 minutes ago, rsvg said:

 

didn't miss any subtle hint. I simply responded to someone else that asked a question after they showed me similar threads. 
 

The thread wasn't 'immediately closed'. A moderator tried to subtly point everyone to the actual feature request thread, when that didn't work, they did it directly.

Now, this feature request has been bumped back to the top and is getting attention again.

 

I would like this feature myself and I hope it gets added eventually. I also understand that the architecture for this software was put in place a very long time ago for a much more narrowly defined purpose.

[roaku 795]

31 minutes ago, byakuya32 said:

Microsoft sql would be nice too rather than a proprietary database engine it has better performance too.

Emby uses SQLlite, which is open source, whereas Microsoft SQL Server is very much not.

Both implement the SQL specification.

[byakuya32 18]

22 minutes ago, roaku said:

Emby uses SQLlite, which is open source, whereas Microsoft SQL Server is very much not.

Both implement the SQL specification.

But Microsoft sql is free so it might as well be

[roaku 795]

2 minutes ago, byakuya32 said:

But Microsoft sql is free so it might as well be

[rsvg 14]

Quote

whereas Microsoft SQL Server is very much not.

Yes, but MySQL is free.. https://www.quora.com/Is-MySQL-free

Edited February 24, 2021 by rsvg

[pwhodges 1527]

5 minutes ago, rsvg said:

Yes, but MySQL is free.. https://www.quora.com/Is-MySQL-free

MariaDB is free, fully compatible with MySQL, and not owned by Oracle!  It's my go-to database these days (after a brief flirtation with PostgreSQL).

Paul

Edited February 24, 2021 by pwhodges

[roaku 795]

6 minutes ago, rsvg said:

Yes, but MySQL is free.. https://www.quora.com/Is-MySQL-free

A user said Microsoft SQL Server was not proprietary.

I responded by pointing out that it is both closed source and proprietary.

I also pointed out that the embedded DB Emby uses is both open source and non-proprietary.

I didn't suggest there were no stand-alone server free and/or open source databases implementing SQL.

Edited February 24, 2021 by roaku

[roaku 795]

2 minutes ago, pwhodges said:

MariaDB is free, fully compatible with MySQL, and not owned by Oracle!  It's my go-to database these days (after a brief flirtation with PostgreSQL).

Paul

Yes, I use MariaDB and OpenJDK.

Boo Oracle. Boo.

[rsvg 14]

12 minutes ago, roaku said:

I didn't suggest there were no stand-alone server free and/or open source databases implementing SQL.

No, but you have been responding for Emby on this thread repeatedly stating how impossible the idea of using a non-embedded database is. I am just pointing out that licensing is not an issue, and further more, that would be the responsibility of the user to procure licensing. 

[Luke 37057]

It's very doable. It's just a question of what we think the demand for this is compared to other features. Given enough demand there's no reason we can't do this.

[roaku 795]

12 minutes ago, rsvg said:

No, but you have been responding for Emby on this thread repeatedly stating how impossible the idea of using a non-embedded database is. I am just pointing out that licensing is not an issue, and further more, that would be the responsibility of the user to procure licensing. 

Oh okay. Do me a favor then. Please don't arbitrarily make up fake versions of things I've said, combine them with unrelated things I have said to other people to create some other thing altogether I also didn't say, and respond to that made up statement instead of the statement of mine you choose to quote since I get confused very easily.

Edited February 24, 2021 by roaku

[ebr 14910]

1 hour ago, rsvg said:

Emby's architecture, on the other hand is so fundamentally bound to an embedded database, that it's developers are completely incapable of enabling the concept of an external database

That's not really fair at all. If you look at the beginning of this thread, both Luke and I state, we actually designed for this possibility originally.  However, that doesn't mean it is just a flip of a switch and we still have to weigh the benefits and the audience that would take advantage of those benefits over the cost.

[rsvg 14]

20 minutes ago, roaku said:

Oh okay. Do me a favor then. Please don't arbitrarily make up fake versions of things I've said, combine them with unrelated things I have said to other people to create some other thing altogether I also didn't say, and respond to that made up statement instead of the statement of mine you choose to quote since I get confused very easily.

You could just be quiet and let us request this feature without failing at attempts to make it sound impossible.. What are you trying to get a free premiere account or something. They can talk to me directly. gtfo of here

[rsvg 14]

9 minutes ago, ebr said:

However, that doesn't mean it is just a flip of a switch and we still have to weigh the benefits and the audience that would take advantage of those benefits over the cost.

Ok, I understand, and as a paying Emby Premiere member hereby request this support formally so that you can add 1 to the list of requests I've seen for this on this forum. thank you.

[roaku 795]

5 minutes ago, rsvg said:

You could just be quiet and let us request this feature without failing at attempts to make it sound impossible.. What are you trying to get a free premiere account or something. They can talk to me directly. gtfo of here

I see you plan to keep making things up. Now that I know it's bad faith and not accidental, I'll leave you to it.

Edited February 24, 2021 by roaku

[shocker 113]

It would be nice to have this implemented. Also with a network db maybe clustering Emby  will be the next big thing

[nayr 36]

I just setup nvidia encoding in my home kubernetes and it was easier than it was getting it working on a desktop, install gpu-operator via helm.. wait a few moments and it self installed drivers on all nodes that had the hardware, then I edited my emby workload to use a gpu resource and that was it.

I deployed a database cluster backend much the same way, few simple commands and full highly available setup deployed across nodes thats self healing.. 

Before K8's this used to the the stuff that took a ton of time to spin up and made no sense for anyone to do at home for anything personal.

Much of the desire here with container orchestration is it scales nicely and once you got a highly available setup going, deploying highly available workloads into it is a piece of cake.. and it makes creating a HA setup with common off the shelf hardware feasible for the home workloads like emby.. Its not so much the performance aspect for me as any one of my nodes is more than enough, its that moving to a home video solution like Emby has put great demands on me to keep it up and online because the very moment it dont work someone's here at my door complaining about it.. so some automated self healing love spread behind my existing load balancers and nodes would be very much appreciated here.. it takes time to detect a problem and spin up new instances, or roll out new changes/upgrades, Ive got to stay up late at night to keep the impact to a minimum when in a K8's cluster I could roll it out w/out interrupting anyone.   

my media is already hosted on NFS server, and mounted to the emby container as read only.. I've got multiple nodes capable of running an emby instance on every node. Its not uncommon for the home lab to have a database server provisioned and tuned specifically for performance. 

I think a common audience here is power users, someone saw enough justification to add in LDAP auth support to Emby then there is plenty of demand for a generic database backend IMO. Let us mad scientists cookup whatever creations in our basements that we want, and few will complain about code licensing. 

Edited February 26, 2021 by nayr

[veehexx1 12]

speaking as a potential new user to emby (currently using MythTV), mysql would be much more attractive than the current built-in/internal method.

it'd also be easier for me to split it away from certain storage types. by default my docker containers go to an SSD BTRFS COW-enabled partition which isnt the happiest with certain sql loads - write amplifications being a main issue. my mariadb instance is on a xfs partition ready to go so having the ability to use something that's ready to go would be far more beneficial and straight forward than having to discover where emby container locates the db and having to --volume mount it elsewhere.

that, the ability to query sql, maintain more granular system backups and i'm sure theres other positives i cant think about... it'd all be good to see.

[MrGrymReaper 0]

On 26/06/2016 at 13:45, ebr said:

 

Yes but that makes another assumption - not just the support of MySql but the support of all the components of our server for simultaneous access to the database by other servers.  That may be implied in your mind but it is an entirely different request (and yet another layer of complexity).

@ebr @dcrdev Even if you don't have multiple servers being able to run off of a proper SQL server as a alternative database store would be grand with a power user, large family and friends access requirements. For instance its also possible for the user to switch out MySQL Server for MariaDB and the version 10.5 release of MariaDB is one sweet SQL server. With features which aren't in the earlier versions or even in MySQL Server, at least some of which would work well in a large setup. Such as being a backend to a HTPC Media Server database data store, its performance alone is impressive along with its reliability and security features.

Edited February 27, 2021 by MrGrymReaper

[ebr 14910]

33 minutes ago, MrGrymReaper said:

Even if you don't have multiple servers being able to run off of a proper SQL server as a alternative database store would be grand with a power user, large family and friends access requirements.

But, are you having an issue that can be directly linked to the internal storage implementation?  That is the question.  Bright, shiny, new faster and technically "better" things are all good but, if there is no real problem to solve, then they amount to significant effort for not much gain.

[rsvg 14]

with all due respect, sql servers are definitely not new, the first one was launched 1989.

I see this as a "no brainer" feature, and the ONLY roadblock for being able to horizontally scale Emby. And yes I am having an issue that is directly related to the internal storage mechanism. I just built a really awesome 3 node ryzen based kubernetes cluster and now I'm going to have to run Emby on my weak ass NAS while all my other services have gained high availability, load balancing, rolling updates, and all the other benefits that only comes with orchestrating containers. I know that I'm not the target demographic, but this feature could be optional perhaps with an env var. 

And I also wanna stress that smaller devices sharing workloads with many containers is clearly the future and it would be a huge win for emby to embrace it. 

Edited February 27, 2021 by rsvg

[ebr 14910]

2 minutes ago, rsvg said:

with all due respect, sql servers are definitely not new, the first one was launched 1989.

And we are currently using a SQL implementation.  You just don't like the particular flavor because it isn't designed as its own "server" (because, in our use-case, it isn't needed by most).

3 minutes ago, rsvg said:

I know that I'm not the target demographic

And that, there, is exactly it.  For now, we need to apply our resources to our target demographic.  Working on this would mean NOT working on other things that many, many more of our users would benefit from than would from this.

I'll say it again.  This is a very interesting item for us and we actually designed for it.  But it isn't a high enough impact feature at this time for us to take resources away from other tasks.