Snowflake Reviews

The primary use case is data platforms, specifically data warehousing. It involves restoring and moving data within the platform to prepare it for analysis, routing activities, or serving as the backbone for applications. 

Snowflake also advertises different workstreams, but my customers mostly use it as their core platform to ingest data and serve the onward goals of the wider company.

The most valuable feature of Snowflake is consumption-based costs, which means that you only pay for the storage and compute you use. There's a complete separation of storage and computing, so you don't need to add another server to increase storage or computing. From a costing perspective, it's well-positioned. 

Snowflake's time travel is also incredibly useful, and they have a function called "UNDROP," where you can undo a table drop. Data sharing and replication for Snowflake are strong, and they have a data marketplace with public and private data sets available for sharing. Companies can put their data on the marketplace, and anyone can use it by starting the payment model. The data is provided live straight to you, and it appears as if it were just another database in your own environment.

The main thing I'm excited to see at some point with Snowflake, hopefully - I've not seen anything coming out of it yet - is Git integration into the worksheets and the UI. Sometimes it can be tricky to manage multiple environments if you're purely using Snowflake as your scripting and pipeline environment. This is handleable, so if you use third-party tools like DBT, Matillion, etc., those can help. But if you're looking purely within Snowflake itself, it'd be great to have some form of Git support.

For the future releases, I would love it if they one day decided to implement their own GUI-based transformation tool environment. I know that many competitors like Azure have to Sign Up, and Azure Data Factory can sit in. However, Azure is a very different beast that serves all sorts of different processes, and an argument could be made for whether it's the best to each of those or not. Specifically within Snowflake, I would love it if they could get some form of orchestration built-in for transformation that doesn't have to be controlled directly through code all the time.

I have been using Snowflake for five years. 

It is an incredibly stable solution. It will only go down if your cloud provider itself goes down. So, let's say your Snowflake is hosted in Azure London. If the Azure London data center goes down, I would only see Snowflake going down. If that does happen, Snowflake does have plenty of options for failback replication and rollover backups. 

So we have quite a few customers that, for example, need their data restored in AWS London, and they've got a backup or a replication stored in Azure London. If AWS London goes down, then Azure London one will kick in and become the primary account, and all of the URLs, etcetera, remain the same because they've set up failover URLs and connections for it. At least for the end customer, there's no change. It's only for the architecture and developers behind the scene who then have to double-check things and do all the normal due diligence. But it runs very smoothly

It is a highly scalable solution. There is no limit on storage or computing. They have everything on consumption-based pricing, but you can have what's known as a multi-cluster warehouse. So, warehouses are what you use for the compute.

The multi-cluster warehouses will sit there originally as a single cluster. But then, if there are enough concurrent queries taking place in that warehouse, it can, as it needs, just spin up another one from another one and another one to meet those current needs. And as soon as they can dive down again, it can switch those clusters off again one by one. And you can create as many clusters, warehouses, as many as you need. There is no scaling issue at all. I've seen it most, like, 10,000 queries a second, and it's run very, very smoothly.

The customer service and support team is very useful and strong. They've got support built directly into the Snowflake UI. So wherever you are on the platform, and you see an issue, you can click into the support area and submit your ticket, including direct things like the query ID that you're using or multiple query IDs and all that stuff. 

I find Snowflake to be very responsive, and if you submit a top-level ticket, you can get a response very quickly. The lowest tier of tickets might take 48 hours sometimes, but overall, they are very helpful.

I personally don't see any of the competing cloud platforms coming close right now to what Snowflake offers. An argument could be made with GCP and Datadog are getting closer. Also, a new AWS Redshift is on the horizon, like a whole new AWS Redshift 2.0. But right now, I've not seen anything that comes close. Snowflake, to my understanding, is the only platform that fully separates your storage and computing, essentially. And it's the only platform I've seen with things like time travel. It's got a whole bunch of great features that I don't know if other tools also have, but it supports semi-structured data. It supports automated tasks, alerts, and reporting. And the data sharing is a massive one. GCP now also has its own data-sharing potential, where you can share data with other GCP accounts. I've not used it myself, but to my knowledge, whilst they have the sharing, they don't have anything that even comes close to the Snowflake data marketplace that allows customers to sell or share their data outside the wider world. And it doesn't have anything that comes close to the kind of private equipment where customers might share their own data internally or to their own. And I think there was one more thing. 

Snowflake also have some really good support for Python, Scalar, and Java through what they call Snowpark, which was launched last year. But more recently, this year, it was announced they're really pushing forward with their StreamLINK integration. It will allow customers to host applications on Snowflake and share those applications with other users in a very similar kind of marketplace environment they use for data sharing. I don't think there's anything that any of the other competitors have right now.

The deployment model is delivered as a service. So the most deployment you have to do yourself is by deciding which cloud provider and region you want it to be hosted in. But Snowflake will actually host it themselves, so there's no deployment beyond clicking from a dropdown and clicking okay, and it'll magically appear.

Moreover, it's very easy to maintain because it's delivered entirely as a service. Snowflake takes care of all the patches, upgrades, maintenance, security tweaks, etc.

We have many long-term customers who have been using Snowflake for years, and they wouldn't continue to use it if they weren't seeing a strong return on investment.

There are many options for starting a Snowflake deployment, but I recommend working with a partner who can provide best practices and guidance. It could be through Snowflake directly or another service partner. Working with a partner can save you time and prevent mistakes down the road.

Overall, I would rate the solution a ten out of ten. 

Our current plan and use case for this solution is to migrate the data from on-premises to the cloud. We are currently using on-prem monitor data and providing it on the cloud. We are using Snowflake so that once the data is in there, we are trying to create shares over it so that external systems can't consume it.

The features I found most valuable with this solution are sharing options and built-in time zone conversion.

I see room for improvement when it comes to credit performance. The other thing I'd like to be improved is the warehouse facility. 

In the next release, I'd like to see easier connectivity to the on-premises tools.

I have been using this solution for about three years.

My impression from using this product for the last three years is that it is quite stable.

My impression is that this is a scalable solution. There are around 200 users of this solution at our company, including three administrators.

The initial setup process was not that complex and it took around six months.

The implementation was done by an in-house team.

My impression is that the pricing of this solution is reasonable.

I would rate this solution a seven, on a scale from one to 10, with one being the worst and 10 being the best.

We're using it more for data warehousing and distribution.

Snowflake is a SaaS platform, so I'm using whatever is the latest version.

It's definitely for compute. The best use case of Snowflake is massive compute. With the parallel reads that we can do from Snowflake, we can combine data from disparate sources, consolidate it, and provide it to end clients through custom stored procedures.

It's user-friendly. It's SQL-driven. The fact that business can also go to this application and query because they know SQL is the biggest factor. So, we can provide all the data, and the analysts, data scientists, and product strategists can go and analyze the data themselves.

Room for improvement would be writebacks. It doesn't support extensively writing back to the database, and it doesn't support web applications effectively. Ultimately, it's a database call, so if we are building web applications using Snowflake, it isn't that effective because there is some turnaround time from the database.

I'd like them to look into the limitations of REST API. Snowflake came up with this native API concept, but it has got a lot of limitations. I'd like to see it provide better service-based APIs so that it can provide data as a service.

I've used Snowflake for over three years.

Its stability is fine, but of late, I get loads of messages saying there's some sort of outage or some sort of issue in the application. I keep getting these notifications from Snowflake, which gives a false impression that something wrong is happening, and it might be underlying in the backend. It doesn't seem that stable.

Its scalability is high. I'd rate it an eight out of ten in terms of scalability.

At this time, we have no plans to increase its usage.

Their support is good.

Prior to Snowflake, it was a completely Greenfield requirement.

It was very straightforward.

It required just two people. One from the Snowflake perspective, and one from my team members' perspective to get the configuration running. That's it.

We haven't yet seen a return on investment because some of the applications are yet to be fruitful and make revenue. We have used Snowflake for the past three years at this point, but we have not yet made great revenue.

It's expensive.

Snowflake is very useful as a data lake and as a data warehouse. Also, it has a lot of features with respect to data science. We are not there yet, but if there are any specific use cases around compute, data distribution, and data sharing, then Snowflake is a tool to be considered.

I'd rate Snowflake a seven out of ten.

We are also using Apigee we have various consumption patterns, data enrichment, and few shedding of the data, and everything goes into Snowflake. If it is multiple consumers, it goes into AMQ, Kafka, or multiple streams to consume. There are specific APIs that we offer after we send the data into the S3 bucket. We have Apigee APIs for consumption, and there are three to four different patterns. For example, we enrich the data, flatten it, and structure everything before the customers going to go into Snowflake. 

There are going to be specific clients who need specific data from the overall data lake, those are going to be exposed as APIs. We have multiple customers needing the same data and for this, we move them into the streaming Kafka.

Apigee does not communicate directly with Snowflake. We have data registration, and everything is coming into something that is called the trusted bucket. The  Apigee interface API is written off the S3 bucket. The S3 bucket data is moved into the Delta Lake, and where the data are stored from the Delta Lake, it sends it to Snowflake. We have Apigee going to Delta Lake and S3 bucket, but  Apigee does not go to Snowflake, these are two areas where it goes to. 

We have Kafka consuming directly off Delta Lake, and it sends data to Kafka through the AMQ. We have its setup, and we have interfaces that come directly to Snowflake to pull the data. It is then flattened and enriched, and it is used for many purposes, such as reporting.

Snowflake's most valuable features are data enrichment and flattening.

I have used Snowflake within the last 12 months.

The complexity of the initial setup of Snowflake depends on the use case. However, Snowflake itself, we don't set it up. The difficulty comes from the ingestion patterns, depending on what data I'm putting in, what kind of enrichment, and what additional value we have to add. However, it does tend to get complex because we have a lot of semi-structured data which we need to handle in Snowflake. There have been some challenges.

Snowflake has multiple implementations. For example, it can be implemented on Amazon AWS and on-premise. The data between these two cannot work together because they have different time zones. That's where the integration can be difficult because it is similar to them being on separate islands, they are completely separate. At some point, everything is going to go into the Amazon AWS Snowflake, but right now there are two islands that are completely different. We have to pull the data out and send it out again separately through a different pipeline.

In the future, this type of implementation should be easier. The integration could be better. 

I rate Snowflake an eight out of ten.

We use Snowflake for our data warehouse. Amazing product. Redshift cannot compete with a true elastic data warehouse where you can scale computing by just issuing a SQL query (increase computer power) and resizing it down or putting computing unit to sleep. 

Snowflake has many more features:

When combined with Alooma, it's the best data integration system. No need for Talend and all these cumbersome tools.

We were able to implement the entire data eco-system in less than five months, from data integration, data warehousing, ELT, producing fact and dimensional tables, and finally reports.

• Computing unit is accessible via SQL: being able to turn them on or off as needed.
• Snowpipe (ingesting data)
• Looking back in time (being able to look at data in the past within a query)
• No data warehouse management
• Support for JSON.

The list is pretty long.

• Snowpipe auto-ingest should be automatic.
• A better client UI or command line tool: I think SnowSQL is a little awkward.

Excellent.

Excellent.

Used it at a previous company.

Yes. No hardware or server config is needed. Just create a user account.

In-house.

Very good.

Snowflake computing is very affordable. Less expensive than Redshift.

Yes. I looked at Redshift and Vertica.

The primary use case is for building a database and data link.

I like the ability to work with a managed service on the cloud and that is easy to start with.

From the documentation, the black box is not very descriptive. Snowflake does not reveal how exactly the data is processed or sourced.

I have been using Snowflake for three years now.

The stability is reliable and a standard product.

The scalability is very good and we have around two hundred data sets currently operating.

Technical support is good. It is readily available and they are very responsive.

Positive

The initial setup was straightforward.

You can do the implementation in-house since it is a managed service and only takes a few hours.

The pricing is economical as compared to traditional solutions like Oracle and competitive pricing.

I would rate Snowflake a nine out of ten.

Snowflake is good for analytical purposes when you have a lot of historical or sales data that you need to release and use for different types of analysis, such as tracking sales and measuring the performance of your sales team and product.

The product's most important feature is unloading data to S3. It provides a single syntax query to analyze data directly from a database to an S3 bucket.

The product's performance could be improved.

I have been using Snowflake for a couple of months. We are using the latest version.

The product is stable.

We did not need to scale our Snowflake environment beyond what we needed. We have a fixed amount of traffic from a fixed number of clients. We know the load we need to handle, and based on that, our subscription is made.

Snowflake is cost-effective. The pricing is better than Firebolt. Firebolt is better when there is idle time. If we run Snowflake all the time, the cost will be higher.

We are working on two solutions for Snowflake, one for the cloud and one for on-premises. It has good documentation. If someone goes through it, they will quickly understand how it works. However, Firebolt's documentation is more comprehensive. If I need faster results, I'll prefer the Firebolt; if I need performance, I'll use Snowflake.

Overall, I rate it a nine out of ten.

We use the solution to build the pipelines in stream sets, including data source, data warehouse, and destination endpoints.

The solution's most valuable features are storage, run time, scalability, and minimum query time compared to other vendors.

The solution's stability needs improvement.

I have been using the solution for seven or eight months.

Recently, I encountered an issue with the solution's data warehouse. The resource monitor had exceeded its quota. I rate its stability as an eight.

I rate the solution's scalability as a nine.

We use Hive and Hadoop as well. Snowflake is more stable and scalable.

The solution is more straightforward to use than the other IDBMS tools. It has a simple query process. Its computing time is less as well. One can easily have access to it. I rate it as a nine.