Someone tried neo4j against titanium - pros and cons

Question

Can anyone point out or point out a good comparison between Neo4j and Titan? One thing I see - in terms of scale - Titan is scalable and requires a basic scalable data store such as cassandra. Neo4j is for HA only and has its own built-in database. Any other pros and cons? Any specific operations. (Is Titan currently used anywhere?)

I also have the following link: http://architects.dzone.com/articles/16-graph-databases-compared , which gives an objective comparison for graphical databases, but does not depend much on the pros and cons between Neo4j and Titan.

Answer 1

We have a social graph in which we add almost 1 million node and twice as many edges per day. We started with the neo4j chart, because yes, it is very fast due to the fact that its storage is located on the same computer that runs the graphics engine. But here is an experience we would like to tell you about neo4j.

• Not suitable for real-time queries. We have a social structure like twitter. We must show the last 20 events (and related activities) of all users whom the user follows his time line. We have several users that are followed by more than 1000 users. In the gremlin request we wrote for this (if you are interested, we can share the gremlin request), in fact, so many GCs were created that the server with 8 cpu and 48 gb ram used to freeze, and we had to restart the server to get it online again.
• There is a multiple network partition.
• There is no vertex center index, which is very necessary in the graoh database.

In the end, we fade away so much with the server performance with a gremlin request that we had to change the database to titanium.

On titanium, we get reasonable performance, and scaling is very simple, since we use cassandra as the backend storage. But remember that .. using gremlin is also not a good idea here, since a multi-task request is very ugly for recording and without multi-git, its request becomes very slow.

Answer 2

It’s great that you study graph databases. I will talk with Neo4j part of your question:

More than 30 of Global 2000s are now using Neo4j for production in a wide range of use cases, many of which are amazing even for us! (And we invented a property graph!)

A partial list of clients can be found below: www.neotechnology.com/customers

Neo4j has been operating in 24x7 production for 10 years, and although the product has, of course, expanded significantly since then, it has been based on a very solid foundation.

Most companies moving to graph databases, speaking for Neo4j, which I know about, do this because either: a) their RDBMS could not cope with the scale and scale of the requirements of related queries; / or b) the huge convenience and speed that arises from the areas of modeling, which are a graph (social, network and data center management, fraud, portfolio, identifier, etc.) in the form of a graph, and not as a table.

For kicks, you can find here several negotiations with clients from four (soon five) GraphConnect conferences that were held this year in the largest cities in the world:

http://watch.neo4j.org/

If you are in London, the latter will be held next week: http://www.graphconnect.com

Below you will find a brief description of some Neo4j technologies with some customer examples. To talk directly about the issue of scaling: Neo4j has a unique architecture designed to maximize query response time and query predictability, allowing horizontal scaling so that each instance can access the graph without jumping over the network. (More reading bandwidth is required. Just add instances.) It turns out that this approach works well for 95 %% graphs, including some production clients who have more than half of their social graphs on Facebook running on the same Neo4j cluster, maintaining the website “always on "24x7.

www.neotechnology.com/neo4j-scales-for-the-enterprise/

One of the world's largest mail delivery services, performs all its packet routing in real time using Neo4j. Railways build routing systems on Neo4j. Some of the world's largest customers use them to manage personnel and data, route alternative routes, manage the network and data center, detect real-time fraud, bioinformatics, etc.

The Neo4j Cypher Query Language is the only declarative query language built specifically for property graphs. He takes all the lessons learned from our 13-year-old native Java API (which was the basis for Blueprints, with which some of the other graphical databases were adopted) and translates them into the next generation language. Cypher is a great way to learn graphics and develop applications; and you always have your own Java API, if you have special needs or a bare metal value (ie submilliseconds versus a single millisecond), above convenience. Neo4j is built from scratch to support graphs and has a graph storage engine that is built to store graphs; unlike some later additions to the ecosystem, graph databases, which are archived as graphic libraries on top of non-graphical databases and are subject to some of their inherent limitations. (For example, FlockDB, since it is based on MySQL, will still be very slow for anything more than one jump.)

Definitely feel free to contact the Neo team if you need anything more specific. We will be happy to assist you! http://info.neotechnology.com/ContactUs.html

Good luck