Is pymongo asyncronous?

Question

I work on a benchmark between pymongo and rethinkdb, to compare the time took for insertions.

However this is what I found :

for one-by-one insertions.


def chronometre_rethink_insert_one(data, nblines):
    avant = time()
    for i in data[:nblines]:
        r.table('test_table').insert(dict(zip(names, i))).run()
    return time()-avant


def chronometre_mongo_insert_one(data, nblines):
    avant = time()
    for i in data[:nblines]:
        db.test_table.insert_one(dict(zip(names, i)))
    return time()-avant

I think that the fact that takes mongo is nearly constant is weird. So I wonder maybe pymongo doesn't insert the data whenever I insert it, but rethinkdb yes, as I call run() on all operations ?

If so, how should I have comparable results ?

user_14852784 · Answer

Is PyMongo asyncronous?

According to their documentation (https://pymongo.readthedocs.io/en/stable/faq.html), PyMongo fully supports Gevent: So yes, if implemented correctly, PyMongo is able to use an asynchronous framework like Gevent.

Details on how what Gevent is, or how to implement it in your example code:

user_10089406 · Answer

It's turn out that I can manage it doing :

client = MongoClient(port=27017, fsync=True)

(adding the fsync), as it "Force the database to fsync all files before returning" (https://api.mongodb.com/python/2.0/api/pymongo/connection.html).

This done, I have result that makes more sense:

Is pymongo asyncronous?

2 Answers