What we talk about far less frequently is our own journey.
One of the reasons CloudZero can help other businesses grow, change, and adapt is because we know what it’s like to shift strategies and try new things that are a little bit scary or unknown.
One such pivot in our history occurred when we decided to begin using Snowflake, our current cloud-based data warehouse of choice. In fact, we’ve been so pleased with this choice that we’ve since joined Snowflake’s partner network.
We felt it was important to write this Snowflake review partly to help our clients understand our history and priorities, but primarily because Snowflake has been a wonderful addition to our stack, and we believe other SaaS companies going through the same things we’ve experienced could benefit from it as well.
Why We Made The Switch: A Snowflake Review By CloudZero
Though Redshift had served our purposes for a while, it eventually became apparent that we needed to make the switch to something else. There were a few areas we felt could be better addressed by a different cloud warehousing solution — we’ll address these below — so started to evaluate other options.
During the evaluation period, when we were shopping around to find the best fit for our needs, we noticed that Snowflake appeared to check all of our boxes. It quickly became the top contender — and our final choice.
A handful of major considerations went into the decision. Specifically, we valued Snowflake’s advantages in three main areas:
Though Redshift operated in the cloud, it still required us to maintain servers to manage it. This was a large drawback because everything within CloudZero was serverless.
Switching to Snowflake provided the perfect opportunity to take our data manipulation and storage serverless as well.
Snowflake handles all the maintenance, upgrades, and other upkeep behind the scenes, so we no longer need servers to handle those things on our end. Since we switched, the reduction in maintenance-related frustrations has been noticeable.
Snowflake offers a rich SQL syntax. We’ve been able to greatly expand the breadth and depth of the operations we use to ingest, build, and otherwise manipulate our data within Snowflake warehouses.
Though it took us a while to fully harness the capabilities of these sophisticated operations, we knew even at the outset that we would want to scale up the complexity with time. Snowflake has allowed us to do exactly that, without any disruptions or steep learning curves.
Snowflake also allows for the convenient scalability of workloads, which was a huge plus for us.
With legacy solutions, you typically would operate a couple of servers, provision them so they can handle your largest workloads, and then pay for those capabilities on a constant basis, whether you use them to the full extent or not.
In contrast, Snowflake allows us to provision resources as needed, scaling up or down as demand rises or falls.
This means we no longer have to pay for resources that sit unused most of the time. Instead, we scale back when demand is low, confident in the knowledge that we can always scale back up when it’s necessary.
This not only saves us money, it also opens the door to opportunities we would have missed out on before.
When a large enterprise wants to work with us, we can now take them on board even if they bring with them ten times as much data as we’re currently set up to handle.
Instead of having to turn them away, we can simply make a few tweaks and set up large Snowflake warehouses that will be more than sufficient to handle the demand.
The benefits of using Snowflake in comparison to a legacy data management solution are so overwhelming, the drawbacks seem paltry in comparison. However, there are a couple of issues that might take new users by surprise.
First is the time it takes to receive a response for certain types of requests.
In Snowflake, when users execute a query, it wakes up a warehouse to run that query. This “waking up” process is necessary because, if warehouses were to run constantly, you’d be paying for them even when they’re not being used.
Most of the time, it takes about ten milliseconds to wake up the warehouse and begin whichever process has been requested.
Every once in a while, however, something hangs in the process and it can take much longer. It might be a minute or two, but occasionally requests appear to pause until the top of the hour. This can be a challenge, but it often seems a small price to pay for the perks of scalability.
The second issue we’ve encountered can be summed up with the famous quote, “With great power comes great responsibility.”
Scaling up to larger warehouses is so convenient and easy, you might find your warehouses ballooning in size before too long. You can always scale back down of course, but many companies struggle to know when to pull back or how to size warehouses appropriately for their needs.
In other words, Snowflake puts so much control in the hands of its users, it’s fairly easy to let costs spin out of control when you’re not paying attention.
Note that this is only a drawback if you have no way to keep track of your Snowflake costs. Thankfully, that’s a problem CloudZero can solve easily.
Is Your Company Using Snowflake? CloudZero Can Help You Drill Down Into Your Costs
Without a specialized cost platform designed for the job, many companies have trouble identifying individual drivers of cloud costs — especially when those costs are obscured under the umbrella of other tools, such as Snowflake.
The good news is that you can harness the power of Snowflake and still understand your cloud costs in fine detail.
CloudZero combines data from Snowflake, AWS, GCP, Azure, Kubernetes, and more so you can see the full picture broken down into granular unit costs.
Achieving cloud cost intelligence for Snowflake means you can build and run your products as efficiently as possible, without worrying too much about unnecessary bloat.