See this issue (https://github.com/orientechnologies/orientdb/issues/4105)
and the associated code (https://github.com/wcraigtrader/ogp/tree/indexes)
for examples of performance with and without indexing edges.

In short, if you're not using indexes on edges, then attempting to find a
specific edge from a given node will involve a linear search through all of
the edges attached to the node.

- Craig -

On Tue, May 12, 2015 at 10:45 AM, <[email protected]> wrote:

> Hi,
>
> Can you please elaborate on this.
>
> As far as I know the edges can not be indexed based on
> outgoing-vertex-properties.
> In this case the edges have no properties and could be light weight.
>
> Regards,
>  -Stefan
>
> Regards,
>  -Stefán
>
> On Tuesday, 12 May 2015 13:14:12 UTC, l.garulli wrote:
>>
>> Hi all,
>> I confirm that indexing the edges is the only way to have a O(1)-O(LogN)
>> time. Once indexed you can start from edges and get in/out vertices.
>>
>>
>> Best Regards,
>>
>> Luca Garulli
>> CEO at Orient Technologies LTD
>> the Company behind OrientDB
>> http://about.me/luca.garulli
>>
>>
>> On 12 May 2015 at 12:47, <[email protected]> wrote:
>>
>>> Hi,
>>>
>>> That is not applicable. Any "where clause" would demand the graph to
>>> visit all edges-vertexes even when looking for the newest 100 (to look up
>>> for the remote property).
>>> When there are 1 million+ edges that takes a very long time and is not
>>> acceptable.
>>>
>>> Thanks for playing!
>>>
>>> regards,
>>>  -Stefan
>>>
>>>
>>> On Tuesday, 12 May 2015 07:54:03 UTC, Izzet Pembeci wrote:
>>>>
>>>> select out('ACTOR')[79220,79221,79222,79223,79224,79225,79226,79227,
>>>> 79228,79229,79230] from #140:0
>>>>
>>>> Can you transform the above query to something like:
>>>>
>>>> select out('ACTOR') from #140:0 WHERE ...
>>>>
>>>>
>>>> If this works in a faster way, one may conclude that implementation of
>>>> [] has some performance problems. I also think that with your reasonable
>>>> bumps you earned yourself the right to open an issue on this one.
>>>>
>>>> iZzeT
>>>>
>>>>
>>>> On Monday, May 11, 2015 at 2:21:27 AM UTC+3, [email protected]
>>>> wrote:
>>>>>
>>>>> Hi,
>>>>>
>>>>> I do believe that in this case neither are applicable.
>>>>> I think the ridbag offers only traversal (not fetching by positions
>>>>> directly) and the edge has not property to index by (it's on the outgoing
>>>>> vertex).
>>>>>
>>>>> Regards,
>>>>>   -Stefan
>>>>>
>>>>> On Sunday, 10 May 2015 15:25:18 UTC, Ziink A wrote:
>>>>>>
>>>>>> I'm still evaluating OrientDB so I might be totally off but I would
>>>>>> index the edge class (SOME_LABEL).
>>>>>>
>>>>>> Also take a look at http://orientdb.com/docs/last/RidBag.html
>>>>>>
>>>>>>
>>>>>> On Thursday, April 30, 2015 at 4:27:40 AM UTC-7,
>>>>>> [email protected] wrote:
>>>>>>>
>>>>>>> Hi,
>>>>>>>
>>>>>>> I have very dense graph that contains vertexes with a lot of edges
>>>>>>> and I need to fetch the X last edges added to the Vertex.
>>>>>>>
>>>>>>> Orient SQL allows me to do it like this:
>>>>>>>
>>>>>>>    - select out('SOME_LABEL')[80000] from #1:0
>>>>>>>    - please note that this is single direction (out) and a single
>>>>>>>    link type / label ('SOME_LABEL')
>>>>>>>
>>>>>>> I have several questions regarding this:
>>>>>>>
>>>>>>>    - Are the edges in a consistent order?
>>>>>>>    - Assuming append-only operations and no deletions
>>>>>>>
>>>>>>>    - Can anything be done to speed this up?
>>>>>>>    - I ask because this query is very slow (0.7 sec.)
>>>>>>>    - Asking for a list "select
>>>>>>>    
>>>>>>> out('ACTOR')[79220,79221,79222,79223,79224,79225,79226,79227,79228,79229,79230]
>>>>>>>    from #140:0" takes almost n*req_time longer
>>>>>>>
>>>>>>>    - What happens underneath (is the whole list iterated from top
>>>>>>>    to get to this)
>>>>>>>
>>>>>>>    - Can this be achieved using the Java API?
>>>>>>>
>>>>>>> Assistance is highly appreciated.
>>>>>>>
>>>>>>> Best regards,
>>>>>>>  -Stefan
>>>>>>>
>>>>>>  --
>>>
>>> ---
>>> 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 [email protected].
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>
>>  --
>
> ---
> 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 [email protected].
> For more options, visit https://groups.google.com/d/optout.
>

-- 

--- 
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 [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to