ByteHouse

I use ByteHouse  to store data, which helps data scientists and data analysts securely store valid data.

Data scientists receive data sets that they analyze, clean, transform, and perform calculations on in order to make the data set valid for machine learning purposes. To store it, we need a platform with high scalability. They use ByteHouse  to store the data.

ByteHouse is an enterprise tool that helps us manage everything easily because all aspects are managed from ByteHouse itself, including maintenance, scalability, and speed. Everything is managed at the back end, so we don't need to worry about it in the model era.

Scalability and cloud-native support are the best features ByteHouse offers in my experience. These are very effective.

For database integration, particularly for DevOps, integrating it with the CI/CD flow became much easier compared to other database solutions. For data scientists performing analysis, it is significantly easier and faster compared with storing data elsewhere.

ByteHouse has positively impacted my organization by making the machine learning life cycle considerably easier and faster. The machine learning life cycle has become much faster because we usually do five deployments per day, and now using ByteHouse, it is possible we can go beyond seven deployments.

ByteHouse is an enterprise solution, so the company has to pay considerably to get it. However, if it were open source, the company could have a trial period and could proceed with confidence to purchase it. Additionally, ByteHouse could still offer a free trial for limited data, such as 1GB or 2GB of data.

Regarding ByteHouse's AI capabilities, some tweaks are still needed. It is still in a growing state from what I have observed.

When considering MLOps or an MLOps life cycle where data sets are stored in ByteHouse, there should be a tool that alerts us when these data sets have missing values, need cleanup, contain duplicate values, or have inconsistent data. Regarding ByteHouse's AI capabilities, I would rate the accuracy as below average, but it is still in a growing state. I would give it a six out of ten.

My advice to others considering ByteHouse is that if you want scalability and your team is quite small, and you want a stable source where you can store your data sets and perform analysis, ByteHouse is a good option.

ByteHouse is stable in my experience.

ByteHouse's scalability is production-grade. It is quite sensitive when it comes to scalability.

We did not avail any customer support because we did not face any issues so far.

I previously used BigQuery , and we switched for the only reason because of scalability and the pricing accordingly.

There is a good return on investment with ByteHouse. We had our own cloud setup and used to store in BigQuery  previously. When we switched to ByteHouse, the price was drastically decreased. We paid for BigQuery previously, and the price was quite higher, ranging from $7,000 to $8,000. When we use ByteHouse, the price has significantly decreased.

We paid $4,000 to $5,000 per month for ByteHouse pricing, setup cost, and licensing.

We evaluated other options before choosing ByteHouse, looking into Dataproc of GCP  and Cloud Spanner. However, we wanted SaaS, so we looked at ByteHouse. Additionally, we received good reviews from ByteHouse.

As I mentioned earlier, ByteHouse is an enterprise solution, not open source. If someone wants to learn it, he will get hands-on experience once he works with it. He cannot gain experience otherwise. If any individuals are interested in learning ByteHouse, they may need to find a corporate organization where it is being used. They cannot directly use it since it is an enterprise solution. From my use cases, I do not find any other improvements needed for data science. However, in different industries, they might have different needs. I would rate this review an eight out of ten.