Thank you!

Em quinta-feira, 16 de abril de 2015 18:49:34 UTC-3, Michael Hunger 
escreveu:
>
>
> Am 16.04.2015 um 21:18 schrieb Daniele El-Jaick <[email protected] 
> <javascript:>>:
>
> Ok, thank you.
> I am just trying to undersatand why a same query, in a same graph, has 
> very different query times in each execution. Sometimes starts with 4000 
> ms, and when I run again decreese to 200 ms. 
>
>
> It's a dynamic system:
>
> - queries are compiled and planend the first time they are run and cached, 
> (that's what you measure in the first run)
>   subsequent queries are taken from that cache and executed directly
>
> - the jvm warms up when the JIT kicks in after a while and optimizes the 
> code
> - garbage collection happens now and then
>
> There will be no fixed time that you get for any query, my suggestion is 
> to run it 5-10 times upfront and then measure the next 10 and take the 
> average.
>
> Also your graphs are tiny and shouldn't make a difference if the queries 
> are sensible.
>
> Michael
>
>
> I am using neo in my doctor degree thesis.
> I need to measure queries times in graphs with 1000, 5000, 1000 and 50000 
> nodes.
> I cant get a precise time.
> Each run has a time, and they are very different.
> In java the times are longer, and graphs with very different sizes have 
> very similar times for the same query.
> I am looking for more precisely form to measure queries times.
> Thnak you again.
>
> Em quarta-feira, 15 de abril de 2015 22:33:08 UTC-3, Michael Hunger 
> escreveu:
>>
>> Try 2.2.1 instead
>>
>> browser roundtrip times, as you return potentially a huge amount of 
>> information also serialization, network transport, json parsing, and 
>> rendering.
>> Michael
>>
>> Am 16.04.2015 um 02:13 schrieb Daniele El-Jaick <[email protected]>:
>>
>> Hi.
>> Under Starter I just found "settings".
>> I´m using version 2.1.5.
>> What affects the execution time of queries in Web Interface to vary like 
>> this?
>> Thanks a lot
>>
>>
>> Em quarta-feira, 15 de abril de 2015 19:19:42 UTC-3, Michael Hunger 
>> escreveu:
>>>
>>> Hi,
>>>
>>> there should be a link to the neo4j-shell under "Options" of the Starter.
>>>
>>> Which version are you using?
>>>
>>> can you provide a query plan with EXPLAIN MATCH ...
>>>
>>> You can also try:
>>>
>>> PLANNER COST MATCH ...
>>>
>>> Michael
>>>
>>> Am 15.04.2015 um 23:25 schrieb Daniele El-Jaick <[email protected]>:
>>>
>>> Hi.
>>> I have installed the "Neo4j Community Installer", which is
>>> .exe packaged Windows installer with bundled JRE. I can´t find Neo Shell 
>>> in it.
>>> I am running several queries.
>>> For example,
>>>
>>>
>>> MATCH p= (e:ENTITY{Cod: 'E9'})-[:INFORMED|USED|DERIVED|GENERATED*]->(n) 
>>> WITH 
>>> DISTINCT pRETURN (extract(n IN nodes(p)| n.Cod)) as Path, (extract(r IN 
>>> relationships(p)|r)) as Rel
>>>
>>> All your paths should be distinct anyway, how many entries are returned?
>>>
>>> You expand your path ad infinitum, e.g if you have 100 rels fan out and 
>>> 10 steps you are at 100^10 paths.
>>>
>>> In a graph with 1.000 nodes this query times vary in web interface from 
>>> 3.600 ms to 140 ms.
>>>
>>> It depends on my internet connection, does not?
>>>
>>> no it does not.
>>>
>>> In java I got 870 ms in average.
>>> Thanks
>>>
>>>
>>>
>>> Em quarta-feira, 15 de abril de 2015 17:31:01 UTC-3, Michael Hunger 
>>> escreveu:
>>>>
>>>> I usually use the neo4j-shell.
>>>>
>>>> It takes a a few runs to warmup the JVM, compile the queries and create 
>>>> a hot-working dataset.
>>>>
>>>> What times are you seeing? And what queries are you running?
>>>>
>>>> Michael
>>>>
>>>> Am 15.04.2015 um 21:47 schrieb Daniele El-Jaick <[email protected]>:
>>>>
>>>> How can I measure query times in web interface? Every time I run the 
>>>> same query I get a different time. 
>>>> In java I get longer times without variation for different sizes of 
>>>> graphs.
>>>> Thanks
>>>>
>>>> -- 
>>>> You received this message because you are subscribed to the Google 
>>>> Groups "Neo4j" 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 "Neo4j" 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 
>> "Neo4j" 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 
> "Neo4j" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to [email protected] <javascript:>.
> For more options, visit https://groups.google.com/d/optout.
>
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"Neo4j" 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