Neo4j AuraDB Professional (pay-as-you-go)

What do you like best about the product?

What I like best about Neo4j is how naturally it models complex relationships, especially for an application like ours that stores interconnected data about arts, artists, places, countries and other entities. In a graph database, nodes represent entities (like artists or artworks) and relationships (like "created" or "exhibited") allow for a highly intuitive representation of how these elements connect.

This makes querying for complex patterns, like finding all artists who influenced a particular art movement or tracing the exhibitions of a certain artwork across different places, efficient and straightforward.

What are the main points that like it more about:
- That Neo4j optimizes queries for traversing relationships, such as "What art pieces were created by artists in a specific location?" which make the response faster than in traditional relational databases.
- We like that you can easily expand the graph with new relationships or attributes as your dataset grows.
- Also, we can search deeper in our data, finding more meaningful connections between our historical data, like trends in art styles or how artists influenced each other across regions, or the several relationship of multiple artist for a specific location or art

The flexibility and performance of graph-based queries really shine when dealing with highly relational data, like historical and cultural information.

What do you dislike about the product?

While the Neo4j offers more positive advantages than disadvantages, but for our case specifically about our history app, there are a few challenges or limitations that might be points of concern, which can be improved:
- First big issue was about the restoring the old data from a different version of the database. Neo4j’s backup and restore processes are more complex compared to traditional relational databases. Maintaining backups for our history app can be a bit challenging, especially with the extensive and interconnected historical data which we are managing. As our dataset grows, ensuring that all this valuable information is securely backed up can require careful planning and additional effort.
- Different query language than traditional ones. Neo4j uses Cypher, which is different than traditional and may require time to learn especially if you're coming from a SQL background like I did. For more complex queries involving relationships between artists, artworks, places, and tags, Cypher syntax can become difficult to manage, especially as the graph structure grows more intricate, you need to optimize the query to not allow a lot of memory time in the whole process results
- Also, one more thing that we find of is importing data into Neo4j, especially from structured sources like Wiki pages, can be more complex than with traditional relational databases. The data needs to be transformed into a graph-friendly format, which can add a layer of complexity when dealing with large-scale imports or frequent updates from sources like Wiki.

What problems is the product solving and how is that benefiting you?

Neo4j Graph Database solves several problems that are particularly beneficial for our history app, which stores interconnected information about arts, artists, places, countries, and types from Wiki. Here are the main keys and topics about how the Neo4j solved our goals:
- First is how efficiently managing big and comples relationships: Neo4j excels at handling complex, highly interconnected data. In our app, each piece of art may be related to multiple artists, places, and historical contexts. Traditional relational databases struggle with deeply nested relationships, often requiring complex joins and leading to slow queries. Neo4j, however, is designed for querying relationships directly, allowing you to quickly find connections between entities like "artworks created by artists in specific places" or "artists influenced by others across time." What is the benefit for our app can offer fast and accurate search results, even with intricate historical data relationships, improving user experience.
- Flexible of the structure for our data: As our dataset grows and evolves day by day, Neo4j allows us to easily expand our graph by adding new nodes (e.g., new artists or art types) or relationships (e.g., "influenced by" or "exhibited at"). In a historical context, new discoveries or data sources (e.g., additional Wiki information) can be easily integrated without restructuring the entire database. The main thing is that the app remains scalable and adaptable, accommodating future data changes without major disruptions.
- Relationships Searching: One thing that Neo4j has ability to search deeper, contextual connections. users might want to explore how specific art movements spread geographically, or how one artist's work related to others across different periods or regions. Neo4j allows us to surface these non-obvious patterns easily, providing richer, more valuable insights to users.
- Performance: As our app will grow up in the amount of stored historical data, maintaining query performance can be challenging. Neo4j is optimized for traversing vast networks of nodes and relationships efficiently, making it ideal for large-scale, relationship-driven queries.

What do you like best about the product?

The ease of creating graph and graph visualisation using neo4j platform

What do you dislike about the product?

The neo4j graph query language is tough for first timer

What problems is the product solving and how is that benefiting you?

We wanted to find root cause of increasing customer complaints for a cpg company's product by connecting the supply chain end to end in graph

Think of Neo4j AuraDB as a special type of database - it's a graph database. Graph databases can be used for situations where you want to do relationship-centric modeling. If you want to identify how data points are related to each other, that's where AuraDB does really well.

Specifically, in terms of RAG and generative AI use cases, where you want to find out how close data points are to each other, AuraDB does really well. It's fast because the data is essentially a graph database with points linked to each other.

It feels like a perfect solution if your use cases involve identifying or working with relationships within the data.

Think of AuraDB as a database. For example, imagine you have textual data in the form of documents, and you want to feed that data into an existing LLM model to gain extra context. That's where you would use AuraDB.

In this use case, you would convert your textual corpus into a graph database and store it in AuraDB. This can then be fed into an existing or newly created LLM model, which will provide better insights. You can then perform analysis on your data, and your LLMs can answer questions and provide better context based on the additional data you've provided.

This is essentially RAG workflow, but it's really useful for storing extra data or storing your data efficiently.

AuraDB effectively manages complex data relationships. If there is an inherent need within your data or your use case to identify how the data is related to each other and how the individual points are related to each other, then the graph structure of the database itself is the biggest feature AuraDB provides.

It also has a query language called Cypher, which is used to query within the database, create the database, and get your use cases out.

So the key features or the key pointers are the Cypher query language, its speed, and the inherent graph structure of the database.

The most beneficial things in terms of AuraDB are its speed, its good pricing, the multi-cloud availability, and its availability across GCP, Azure, and Amazon. It's great for use cases where you want to do relationship-centric modeling. So, those are the most valuable things in AuraDB.

I also work with real-time data in the AuraDB solution. A lot of this, especially the scalability and how efficient these conversations are, depends on what model or writing strategy you go for. But you can definitely work with real-time data.

For my personal projects, I use AI. What we're seeing right now can work very well with RAGs in AuraDB or any graph database. So we take extra data, put it in a graph database—AuraDB in this case—and feed it to an existing large language or a small language model. With that, an AI model can gain some extra understanding of your data, which is stored in a graph database.

It can give out very contextual and specific answers based on the extra data users provide in the form of a graph database, which is stored in AuraDB. So the use cases are, from what I mean, the terminology is graph RAG, but that's where I see a lot of potential use cases for a lot of data.

The outcome accuracy with the AI-enhanced graph is good for my use cases. However, it may be difficult to assign a numerical accuracy metric to Neo4j. But for example, with text summarization, you cannot put a number to the accuracy. However, just seeing the answers and the improvements in the model, it's definitely helpful in improving the results. It's essentially giving an extra context to your model. So, I definitely see the advantages of using AuraDB.

I've been using it for a few months now, and everything has been fairly positive. Maybe in terms of documentation, they can improve a little bit. 

Neo4j AuraDB already has a good set of documentation, and the initial setup is easy, but it could be made a bit easier. For me, things are going very well, actually.  

In terms of AuraDB, the conversations have always been around scalability. So that's where people are majorly concerned: whether it can be used for truly production-grade projects. But Neo4j AuraDB consistently comes up with updates. But potentially, that could be one area where maybe I can see some more improvements.

I have been working with AuraDB for around six months now. It's mostly been an experimental thing where I try out projects and find use cases to see its maximum potential.

I do find it stable. There are some competitors out there, but in terms of the learning curve, it's very easy. The initial setup is very easy. So, it's definitely a stable solution.

Five years back, scalability was considered a bit of an issue with respect to AuraDB. But I think with the recent updates, they've handled it very well.

Currently, I'm using AuraDB just for experimental purposes, so from what I've read and what I've seen about AuraDB, it can handle quite a vast amount of data.

There may be some performance issues when your database or your data is very large, but then again, it's completely dependent on what pricing strategy you go for.

From my side, right now, it has been mostly experimental and working on personal projects. So, again, it's dependent on what project I've seen. But it can also be used for large-scale projects. That's where I see conversations where people are a little bit concerned, wherein very large use cases, where billions of data points are there, whether it would be as efficient. It would work, but maybe it might take a hit in terms of speed, even the efficiency of it.

As of now, I have not reached out to them as such because everything has been fairly clear to me. But I'm fairly sure that the technical support is good.

I have not worked with other graph databases, but I am aware of the competitors. There is TigerGraph database, and I think Amazon Neptune, and one from Azure as well. I've not really worked them out, so I use AuraDB.

I found the initial setup fairly straightforward. From what I felt, the learning curve was a bit simpler. AuraDB had their courses out there, and some of them are out there for free, so you can just quickly learn them. And I just felt that the initial setup was much simpler compared to others, and I was able to catch on to it. 

The deployment is just a standard way—it's like any other database. There's no difference in the way AuraDB does things. 

AuraDB can be hosted or is available in the major cloud services. So, the deployment procedure remains pretty standard compared to the other existing databases out there. There's no difference as such.

We use the public cloud, so that's where the deployment is being worked out.

The deployment time depends, again, on the project and the circumstances. But, the initial learning, it might take two to three months to pick it up. And working on a project, again, maybe another three, four months. And in terms of deployment, another one, two months to it. But, again, it's purely dependent on the project and the circumstances.

From what I have seen, there's no real maintenance or anything extra to it. It's just that since it's a new technology, or rather, not many people might be aware of it, it's just the awareness needs to be there, but there's no additional maintenance as such.

I have done the deployment myself. There has been no real assistance, at least until now. But I think their community support is fairly nice, so that's something to look out for as well.

The product offers three pricing strategies. 

One is the free version of AuraDB, which can be used for small and experimental projects, which is what I'm doing. 

Then there is AuraDB Professional, which is $65 a month. 

And then there is AuraDB Enterprise, which is for the production of large-scale use cases, and that's where they give more security and support. 

So those are the pricing strategies. 

I use the free version as well.

I would definitely recommend AuraDB to others. Give it a shot to see whether it fits your use case, and I would definitely recommend it.

So, for my current usage, I would give AuraDB a nine out of ten. I think it's fairly good. Again, the small improvements might be in terms of the scalability and a little bit more documentation, but a fairly solid nine out of ten.

Public Cloud

I worked on a project focused on the quality of public menus, using Neo4j AuraDB to connect and create relationships between food items. This allowed us to visualize data in interesting ways and identify communities. A key feature was using the Green Dot to link unstructured data, such as investment information, with structured data from tables and PDFs. The AuraDB documentation was also helpful in making these connections and providing valuable insights.

The most valuable features of Neo4j AuraDB include its flexible data model and broad language support. It’s great that it offers a dedicated query language, which delivers excellent performance and high availability. Additionally, it’s hosted on AWS Cloud, which ensures reliability. The platform also allows for the integration of videos and other media.

Some features can help if they can visualize graphs better.

They have Neo4j Bloom, which is great for visualization. If you can visualize the graph directly within Neo4j AuraDB, that would also work well.

I don't have any problems about the performance

Scalability is very good.

I’ve used RDP before but prefer to start my analysis with Python and sometimes Neo4j Bloom. The most important feature is that Neo4j is a powerful graph database, enabling faster and more efficient analysis.

It's very simple to create a cloud account, and it takes a few minutes to deploy.

ROI is nice because you can have an incredible return.

It has fair pricing.

The community is very nice, and you can find many things.

Neo4j AuraDB is a powerful graph database that enables us to accomplish impressive tasks. Specifically, as a cloud-based service, it eliminates the need for a high-performance computer to use it.

Sometimes, I collaborate with Smiths when working with large amounts of information. To streamline the process, I often use a chatbot agent plugin, which allows me to respond quickly in real-time, improving the overall user experience.

I've been using this chatbot agent for investment-related projects, but my first project focused on maintenance and public school menus. This initial project is more important because it involves public schools, children, and food insecurity. Conducting this analysis and developing the AI project with Neo4j could lead to meaningful results in the future. We can improve the accuracy of the model by providing context. I can't supply the necessary context if I use traditional methods, like vector regression. However, by creating a knowledge graph in Neo4j AuraDB, I can offer this context to the model, leading to better accuracy and performance.

It's very easy to maintain it.

It's an incredible tool that is quick to use and delivers impressive results. Many people should give Neo4j AuraDB a try. It's a very effective graph database.