How to speed up "FIND REFERENCES #64:0" SQL

[Hung Tran]

Hi,

I am running OrientDB version 2.1.8, from Studio, it took long.

it took more than 2 seconds

Here is its explain

It takes the same amount of time on next attempts.

Any idea will be really appreciated!

My Best,

Hung Tran

[alessand...@gmail.com]

Hi,

how many record do you have in your database ?

Kind regards,

Alessandro

[Hung Tran]

Hi Alessandro,

For the cluster 64, there is just a single record and no dependency.

For whole database, there are less than 110k records

Is there any clue?

My Best,

Hung Tran

[alessand...@gmail.com]

Hi ,

I think it depends by the number of records in your database.

Best regards,

Alessandro

[Hung Tran]

Hi Alesandro,

so that is really a serious issue. A database with few hundred thousand of records are just a pretty small database in a business application.

Is there any workaround for this?

My Best,

Hung Tran 

[scott molinari]

What do you need that command for?

Scott

[Hung Tran]

Hi Scott,

I have two use cases for this

1. Because OrientDB allows you delete any record even if they got referenced by some other records. I need to check its references to know when a record could be deleted, when it could be not and do a soft delete instead, and proceeds all related records according to ORM model. For example, with Sales model, when you delete an draft order, you will need also to delete its all order lines but not delete any Product referenced by them.

2. I have a special use case, when a record is updated, it will do an update only if that is not being referenced by others, otherwise it will do an insert instead.

Without that SQL, I will have to generate many SQL to compute its dependencies.

My Best,

Hung Tran

[Luigi Dell'Aquila]

Hi guys,

FIND REFERENCES command is not intended to be used in typical production environments.

Links are unidirectional, so to find all the documents that refer to a single RID, OrientDB has to do a full scan of the database, this is operation will take a long time to execute.

We suggest to use FIND REFERENCES only to repair databases where you deliberately deleted records without updating their references.

For normal use cases we suggest to:

- use edges instead of links: edges are bi-directional (no need to execute FIND REFERENCES) and are automatically managed by OrientDB (they are automatically removed when you remove a connected vertex)

- use links only in schemaful, indexed properties, so that you always know where to look for in case you manually remove a linked document

There are no plans to make FIND REFERENCES faster, just because it would imply a huge performance impact on the whole architecture

Thanks

Luigi

2016-01-21 3:55 GMT+01:00 Hung Tran <tdhu...@gmail.com>:

Hi Scott,

I have two use cases for this

1. Because OrientDB allows you delete any record even if they got referenced by some other records. I need to check its references to know when a record could be deleted, when it could be not and do a soft delete instead, and all related records according to ORM model. For example, with Sales model, when you delete an draft order, you will need also to delete its all order lines but not delete any Product referenced by them.

2. I have a special use case, when a record is updated, it will do an update only if that is not being referenced by others, otherwise it will do an insert instead.

Without that SQL, I will have to generate many SQL to compute its dependencies.

My Best,

Hung Tran

On Thursday, January 21, 2016 at 2:01:51 AM UTC+7, scott molinari wrote:

What do you need that command for?

Scott

--

---
You received this message because you are subscribed to the Google Groups "OrientDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email to orient-databa...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.