I'm struggling with Neo4j Browser. I have loaded Movies database into an Enterprise Single
Edition 4.0.3 installation on an r4.large EC2 instance.
It is a nice configuration:
2 cores, 16G RAM, 70G HD
And there is only me accessing it.
Even so, it is slow, very slow.. a really poor performance for a single and tiny query:
MATCH (people:Person) RETURN people LIMIT 10
It took almost 1m15s to perform. More of this, drag or double clicks, zoom in/out operations take more than 30s to perform. At least.
So I read the log, there is nothing there and I think the problem is the Browser app, although I am not sure.
What can I do to improve this?
In general the Neo4J browser is clunky and slow. So I wouldn't rely on it for graph exploration.
Yes, you have a point. However, when I was using a sandbox instance it was normal, an acceptable response for every action I take there. I don't understand why got so slow with a dedicated server.
Did you try doing the same query using
cypher-shell? I would first compare the speeds on there using the
PROFILE prefix to understand whether it's the browser / network problem or if there's an issue with the performance of your database.
No, I didn't. I know
cypher-shell, this is a great idea but I don't know
PROFILE prefix. How does it work? Can you send me a sample?
The same thing as a normal query, you just put
PROFILE in front, e.g.
PROFILE MATCH (n:Node)... and it prints out some extra details as to how many seconds it takes to perform your query, how many db hits it made, etc.
What browser are you using? Chrome? Safari? Edge?
@mckenzma — It is pretty well-known that Neo4J browser works very badly in all those browsers as soon as you're dealing with anything bigger than a few nodes. I've been watching this since 2013 when I first started using Neo4J and it's amazing that in 7 years it's still as slow as it was back then. I always thought that maybe it's a special trick to let the other companies do the job, so maybe you know which alternatives could be used?
Something seems off. I have Neo4j Browser version 4.0.3 running for a db I work on with billions of nodes and can return a few hundred without issue.
@mckenzma that's pretty amazing. maybe the reason is that my datastructure has a lot of connections. so even if the number of the nodes is low, the number of connections is high (that's because each connection is made by a different user in a different context and so that is accounted for). could that be an issue perhaps?
Even with a ton of connections, the length of time your queries takes to run depends on many factors. If you have some code examples, you can paste them and get feedback.
Additionally, depending on the size of your graph, you may want to adjust your memory allocations in the config to increase your ability to read through and return data.
So far I am using Chrome and Opera, mostly the first one
And it is the simple
MATCH statement listed above that is causing issues?
Yes, is taking an absurd amount of time to make a drag & drop operation, a double-click operation and for this match takes more than 1m30s to start executing.
That is very atypical in my experience, what kind of connection is there between your computer and that AWS instance? Is there any lag when you try to ping the IP of your instance?
It is a normal connection, bolt or http. I have secured the EC2 instance through security groups, and AMI is one released by David Allen for EC2 environments.
So far I restarted the db server, the EC2 instance, checked all logs and everything looks normal.
Does it also happen when you return properties (n) ?
Please also try cypher-shell
I've had the same problem the last few weeks using an EC2 instance through Chrome. I tried connecting through the desktop app and everything worked well, and I also tried through the new chromium Edge browser (first time I'd really used it) and it worked really well there as well.
Nice. And now, it is working with Chrome or you are accessing via Edge?