Oak with MySql for large scale system

[yogesh upadhyay <up...@gmail.com>]

Hello,


We are planning to use Oak with MySql and building custom authoring
interface for CMS. We are expecting to have millions of nodes and high QPS.
Unfortunately I couldn't find any benchmark or performance test results for
Oak with MySql. Can any one help us to find it (if it exist). Also is there
is any thing we should aware of when dealing with large amount of data
(except flat vs hierarchical data, ordered vs unordered nodes) ? For
example, I couldn't find any document that talks about data store garbase
collection, revision cleanup etc for oak with MySql. I could not find any
document or recent bug related to data corruption and steps to resolve them.

Also with MySQL now natively supporting JSON data type,
https://dev.mysql.com/doc/refman/8.0/en/json.html is there is any plan to
change how data is managed in MySQL (table vs Json data type).

Any help is appreciated.

Yogesh

Re: Oak with MySql for large scale system

[Julian Reschke <ju...@gmx.de>]

On 9/30/2018 1:00 AM, yogesh upadhyay wrote:
> Hello,
> 
> 
> We are planning to use Oak with MySql and building custom authoring
> interface for CMS. We are expecting to have millions of nodes and high QPS.
> Unfortunately I couldn't find any benchmark or performance test results for
> Oak with MySql. Can any one help us to find it (if it exist). Also is there
> is any thing we should aware of when dealing with large amount of data
> (except flat vs hierarchical data, ordered vs unordered nodes) ? For
> example, I couldn't find any document that talks about data store garbase
> collection, revision cleanup etc for oak with MySql. I could not find any
> document or recent bug related to data corruption and steps to resolve them.

Garbage collection is essentially the same as for MongoDocumentStore.

>  Also with MySQL now natively supporting JSON data type
> https://dev.mysql.com/doc/refman/8.0/en/json.html is there is any plan to
> change how data is managed in MySQL (table vs Json data type).

Currently there are no plans to do that (but hey, it's open source...).

Best regards, Julian