Snowflake’s website shows a matrix of editions and clouds. How do I pick what’s right for me?
About the choice of infrastructure platform.
First, let’s get something out of the way. Snowflake is a SAAS (“software-as-a-service”) data platform. It leverages unique properties it sources from public cloud infrastructure. That means that companies cannot install Snowflake in their own, physical, on-prem datacenters as these don’t have the fundamental services such as S3, EC2, or Blob Store that Snowflake uses internally.
Regardless of the deployment option of choice, data platform architects need to work with at least one of the 3 major cloud platforms that are currently supported:
- Amazon Web Services
- Google Cloud
- Microsoft Azure
As such, Snowflake is available in a mono-cloud strategy, i.e. by having its technology and all potentially auxiliary components altogether bounded by the same infrastructure cloud provider.
So I picked cloud X. How do I exactly deploy Snowflake?
Snowflake works typically in a shared tenant architecture. Meaning that they purchase capacity from a cloud provider, deploy and maintain their own software on that cloud provider’s account, and resell access to their software. The company doesn’t charge for the sum of your individual cloud charges, but the price options cover all underlying component charges.
In case a trial account backed by a credit card wouldn’t fit your use case, get in touch with us as a reseller to have an invoice-backed Snowflake account set up.
As such, deploying Snowflake is as easy as filling in a registration form on their website to open up a trial account. An automated process on Snowflake’s side will deploy your own isolated tenant in their shared infrastructure, and you’ll get access credentials in your mailbox.
But what if I don’t want to choose a single infrastructure provider?
Assume you have applications or application components spread across 2 or 3 clouds, and want to optimize compliance, contracts, or a general vendor strategy. Then you’d be able to deploy Snowflake to any of these infrastructure clouds. You would optimize security and cost by doing so. And your data can be easily replicated between all deployments, making sure there are no data pipelines or migration efforts required.
Picking a Snowflake edition
The Data Cloud comes in various editions, actually a combination of features and contract terms that make up a service level agreement.
Here’s how we decide over the various editions
Snowflake Standard edition: great for non-production workloads
This is the base edition, meant for safe experimentation and getting a flavor of what the data cloud could bring. It is useful for small scale deployments that don’t have strict requirements in terms of SLA management on data products. Unique production-ready features that make Snowflake worthwhile are somewhat limited
Snowflake Enterprise edition: great for machine data
The enterprise edition covers the SLA’s business needs to run production-ready workloads on any volume of data. It comes with time-travel capabilities that well extend the requirements of many business processes.
Snowflake Business Critical edition: great for sensitive data
Some use cases require more in terms of contract terms, uptime SLA or shielding access to data. The business-critical edition extends the Enterprise edition with “hold your own key” encryption capabilities and secure traffic routing that is not interceptable over the public internet.
Virtual Private Snowflake
Security and compliance may have approved the use of a certain infrastructure provider. However, they might not allow the use of a shared tenant infrastructure. That’s where Virtual Private Snowflake comes in. It has the capabilities of the Snowflake Business-critical edition. To cover compliance needs, VPS is available on an infrastructure cloud account that is owned and governed by your organization. There won’t be any other outside tenants on that account, improving security even more.
Snowflake community edition
Many SAAS products – often originating from the open-source world, offer a free-to-use model with limited capabilities in their SAAS portfolio. Snowflake currently does not. As an alternative, they offer free time-constrained trial accounts.
Deploying the Snowflake tenant by itself gives you access to the technology. This is only a starting point. Your cloud engineering team still needs to configure Snowflake to access your enterprise data in the most secure way. Depending on your workloads, you might have to design and configure extra – outside – components.
To avoid overhead architecture, we have packaged over 150 best practices to configure these integrations in a quickstart package.
Reach out for more information, and meanwhile, good luck in picking your Snowflake deployment options!