MongoDB Atlas for Government

What do you like best about the product?

Very fast, network-efficient, and affordable. It’s easy to use and integrates smoothly with existing applications and APIs. The developer documentation is clear, and support is readily available when needed. I also like the added AI features for optimal data analysis and further research on my data.

What do you dislike about the product?

The way data is segregated creates too much fragmentation, which makes it hard to switch between projects, clusters, databases, and collections when I’m trying to create data or simply find what I need.

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

I run multiple high-performance database workloads on MongoDB Atlas instances so our customers can query and manipulate large volumes of data from our telco records.

What do you like best about the product?

The most useful thing I’ve found is actually a very basic feature, but it has genuinely saved me. When I was performing a migration in a client environment at my company, I had to constantly keep track of which database/connection I was currently working in. MongoDB Atlas lets you color-code connections, so I can immediately tell, “Okay, I’m in this DB/this connection,” without second-guessing myself.

Another thing I like about MongoDB Atlas is how easy it is to use with Node.js, Python, Golang, and other languages. They also provide automatic monitoring, so I can see things like CPU usage, and there’s autoscaling in place. That means I don’t have to worry as much about whether I’m handling the right number of users or if the setup is sized properly.

One more point: my company uses both AlloyDB and MongoDB, so we have data in both places. Because of that, it’s been pretty easy to integrate and work across them. I also work in an MNC where we build a lot of AI agents, and MongoDB’s features for chunk embeddings and vector search have been helpful. I believe those are in the paid version, not the free one, but using them reduced a lot of back-and-forth between databases for us, since MongoDB can effectively work as a vector DB too.

On pricing, I do have some issues. We were using the VertexAI integration, but now we can see that the Voyage AI integration is cheaper in MongoDB, so we have to migrate. While migrating, I will say it’s convenient that there are a lot of commands I can run to move data directly, like mongodump and similar tools.

As for support, I’ve never really faced issues that required it, so I haven’t had the chance to contact the Atlas team. In my case, that’s usually handled by our DevOps person anyway.

What do you dislike about the product?

I didn’t get the free vector DB integration I was expecting. I’d really like some kind of educational scheme where I can use the vector DB without paying, but that doesn’t seem to exist right now, and that’s a downside for me.

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

A big problem we were facing was how to store our data in a way that stayed flexible as our product evolved. Our company builds AI agents, and we create a lot of them. The structure we work with is something like this: an assistant can contain multiple agents, each agent can contain multiple tools, and each tool can have integrations. Each integration stores the data specific to that integration.

To give a simple, superficial example, imagine an assistant called “human.” Inside “human,” the tools might include a hand (maybe only a hand, or maybe hands and legs). At the top level, the assistant can also have its own properties—for example, if it’s a human, it might have a property like “anger.” Then, within the hand tool, you’d have the hand’s properties, plus its integrations. Even within those integrations, you might have different variations—for instance, one integration could represent fingernails, another could be a gauntlet, and another could be gloves.

The point is that we needed a very flexible schema. If we had used another database, it likely wouldn’t have supported this kind of structure as easily, and we wouldn’t have been able to model things in such a nested, adaptable way. With Mongo DB, we were able to do it.

Another thing I want to add is that since Mongo DB introduced transactions, they’ve been really useful for us. If a transaction fails, it automatically rolls back, and the whole step fails cleanly, which is a great feature.

What do you like best about the product?

I like the speed of response that MongoDB Atlas offers. The clients need faster response to interact with chats, and MongoDB Atlas works well in this aspect. Switching from Postgres to MongoDB Atlas also simplified the process of database maintenance for me.

What do you dislike about the product?

I find that MongoDB Atlas could use more integration with Rust, which is a bit of a limitation for my needs. Additionally, having more documentation available and resolving current problems with Rust integrations would significantly improve my experience with the platform.

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

MongoDB Atlas simplifies maintaining databases and delivers faster responses for client interactions in chat applications.

What do you like best about the product?

Performance Advisor provides helpful insights into query performance. The index usage information also offers a clear view of how our application code is using indexes, which makes it easier to understand and improve overall performance.

What do you dislike about the product?

DISK IOPS is not clear because it depends on various backend infrastructure. I have experience with AWS ENI, and in my case it wasn’t obvious how the DISK IOPS would behave or what to expect.

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

I manage a large Mongo cluster with around 40 TB of data. Performing maintenance used to be a hard task. If I needed to increase disk space, I had to expand the disk and rebalance RAID across my volumes in EC2. That has been resolved after migrating to Atlas. I enabled auto-scaling for storage, and I also enabled online archives.

What do you like best about the product?

MCP integration. Multi-Tenant Architecture. Ability to seamlessly increase the size of the cluster. The Performance Advisor is also a great feature.

What do you dislike about the product?

Sharding doesn’t feel very intuitive to me. I have to stick to a single shard to prevent elements from reappearing.

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

Scaling, reliability, and implementation.