What is the best graph db pooling mechanism to use when scaling to
thousands of Orient database users? I found a few forum threads on this
using OPartitionedDatabasePoolFactory , but I could not find any mention of
it in the user docs - so wanted to confirm the usage. I'm also assuming
that this pool is a must when using Orient db-level security that depends
on having user specific connections (as opposed to just a generic admin
user to do everything)
Based on the javadoc, as far as I can tell, this is the usage pattern:
//One time
//Keep a single static instance of the pool factory. Default LRU cache size
is 100
//Pools may be evicted when capacity is used
private static OPartitionedDatabasePoolFactory PDF = new
OPartitionedDatabasePoolFactory();
PDF.setMaxPoolSize(...); //All pools from this factory will have this max
pool size (default is 64)
//Everytime the user does something.
//Depending on concurrency, thousands of these pools will get
created/evicted.
try {
//Pool for this db/user
OPartitionedDatabasePool
<http://orientdb.com/javadoc/latest/com/orientechnologies/orient/core/db/OPartitionedDatabasePool.html>
pool
= PDF.get("remote:host/dbname", "uname", "upwd");
//Get a db connection for this thread
ODatabaseDocumentTx db = pool.acquire();
//We need an Orient graph, so wrap it. This is a lightweight operation
OrientGraph g = new OrientGraph(db, true);
//Do cool stuff with g (tx is auto start)
g.commit();
} catch (Exception e) {
g.rollback();
throw e;
} finally {
//Return the connection to the pool. Is this right?
g.shutdown(); //This is confusingly called shutdown, but apparently just
returns to pool? Is this right?
}
//One time on application shutdown.
PDF.close();
--
Question: Is this usage pattern correct, and will it scale to thousands of
users?
I based the above usage on what I found in an older thread reproduced below:
"But problem with creation of OrientGrah instances with many users have
already been solved by us .You see OrientGraph is merely thin wrapper on
ODatabaseDocumetTX instance , creation of which I agree is expensive task .
To solve problem when you have millions of users in database we have
created OPartitionedDatabasePoolFactory which is lock free LRU list of
connection pools for each user , each connection pool not only lock free
but for many classes wait free because it tries to partition load between
active CPU cores ( indirectly of course ) . According to our benchmarks
acquiring of connection from this pool requires nanoseconds . When you
acquire database instance you merely wrap it in OrientGraph instance which
is cheap operation"
Thanks,
--Harish.
--
---
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.
