On 17.06.2024 12:17, Raffaele Gambelli wrote:
Good morning,
I would like to ask the community for some advice and information, in the
company where I work we have been developing for years a product based on
jackrabbit 2.20.13, it is a software that allows you to model data according to
any need, so each installation has distinct data types in addition to a common
set.
Any data type is managed through jackrabbit, and any property is indexed to
allow any type of search.
We have always used rdb-type persistence (using both oracle and postgresql and
now mariadb) with a file-type datastore (some installations have as many as 10
tera datastores).
In some situations we have also created a jackrabbit cluster.
We have also used the concept of multitenancy by implementing it as separate
repositories with respective workspaces
We are starting to figure out how to move to oak, it is likely that I will come
back to ask more questions to understand more how others have exploited it.
First, I would like to ask if there are any best practices on the choice of
node storage and blob storage type?
AFAIU, segment store has higher performance, but document store can be
clustered.
From what I understand, for node storage you advise against using mongodb in a
production environment by favoring the use of rdb-type persistence, is that
correct?
No. MongoDB works just fine in production. The RDB implementation
essentially emulates the MongoDB Java SPI on top of relational
databases, so is unlikely to perform better. Furthermore, it hasn't
received a lot of attention in the last few years, so you should plan to
run extensive tests, and maybe have to do a bit of work to align the
code with the current MongoDB support.
the use of a file datastore is great for large volumes of data, is that correct?
I cannot understand the difference between FileBlobStore and FileDataStore, I
only understand that the latter is ideal when migrating from jackrabbit 2.x and
also perfect for sharing between multiple repositories, is this correct?
I'm no expert here, but you'd have to choose the data store in a way
that it works for your setup (clustered?).
Are there possibilities to encrypt node and blob data? For nodes I suppose it
depends on the type of node storage, for example with mariadb encrypting data
on the database is easy, for blobs would you know how to recommend something?
Thank you very much
...
Best regards, Julian
