We uploaded the database online, in case anybody wanna run tests: http://orientdb.com/public-databases/pokec.zip (1,2Gb)
Best Regards, Luca Garulli CEO at Orient Technologies LTD the Company behind OrientDB <http://orientdb.com> On 22 June 2015 at 00:06, Luca Garulli <[email protected]> wrote: > Hey guys, > We sent the PR, but seems it takes a lot of time to merge it and re-run > tests... > > However, Enrico (maggiolo00) took a look a this benchmark and optimized > the OrientDB implementation in the following ways: > > - Using OrientDB 2.1-rc4 instead of 2.0.x. This seemed more fair since > the release from Arango was the last alpha > - The database was created without lightweight edges. Once we > re-imported it (by the way we used OrientDB ETL and a couple of JSON files) > everything was much faster > - The singleWrite test used SQL statement, but this is not the fastest > way, so Enrico used direct document create. On this test we are the fastest > DBMS on insertion > > Everybody can clone this repository and run the benchmark by himself ;-) > > https://github.com/maggiolo00/nosql-tests > > And this is the only commit Enrico did: > https://github.com/maggiolo00/nosql-tests/commit/b8fdddf9662322d748dfdbc5dd18787db2b75416 > > However since on these days we adopted the OrientJS driver from Oriento > project, we found some issues and bottlenecks, that on this tests made the > difference. For example on "singleRead", OrientDB was very slow. On my PC > it took about 60 secs to execute 100k of reads, but once we profiled the > time spent we discovered that 70% of the time was on Node.js driver and > only 30% to execute the query! While with marshalling, the Node.js driver > is good enough, with unmarshalling it's very slow. For this reason we're > going to improve the unmarshalling in the next weeks. Stay tuned to get the > updates on this. > > NOTE: it's not that the OrientJS (forked from Oriento project) > implementation was bad, we think Node.js is not so good to manipulate > chars/bytes, so we're considering new options to do that ;-) > > We run the benchmark multiple times and OrientDB was the fastest DBMS of > all the others tested, but "singleRead" (see above) and "neighbors2" > (actually on "neighbors" OrientDB is the fastest). By looking at the kind > of benchmark we understood why: it's not what you can expect by a classic > 2-nd level neighbors, but it returns only the IDs. On Arango, like any > other Relational DBMS, you have primary keys that are on indexes. > > So that particular query uses the index without even fetching the real > documents. That's why seems faster, but retrieving the ids is an edge case, > without any particular meaning in a real use case. If you're looking for > neighbors you usually are interested on any information about the > neighbors, like name, city, etc. Not just the IDs. > > However, even if this benchmark has been created by a vendor to > demonstrate that is the fastest, the complexity of the benchmark is very > simple. So I'd call it micro-benchmark. Furthermore other vendors weren't > called to do any tuning, so I see this as a mere marketing move to make a > lot of noise. > > > Best Regards, > > Luca Garulli > CEO at Orient Technologies LTD > the Company behind OrientDB <http://orientdb.com> > > On 21 June 2015 at 23:33, Marvin Froeder <[email protected]> wrote: > >> I would love too see this PR... Just to make sure I'm not doing the same >> mistakes :) >> >> -- >> >> --- >> 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.
