Naveen Zutshi is the chief information officer of Databricks. Previously, he was the chief information officer of Palo Alto Networks.
As budgets tighten, leaders are feeling the pressure to temper spending, and tech teams are no exception. Many companies invested heavily in the cloud during the pandemic and are now looking at ways to optimize resources.
According to a recent report, as many as 81% of IT leaders have been directed by their C-suite to reduce or halt cloud spending, which represents 30% of IT budgets. It’s a critical moment for CIOs and other technical leaders to take stock of their cloud and IT budget and usage.
While it is not advisable to abandon cloud-first strategies in favor of on-prem or hybrid infrastructure, it is possible to reduce cloud spending significantly. Databricks recently cut our total cloud spend by 25%, and we’re tracking to reduce total SaaS IT spending by 30%. We accomplished this by democratizing the cloud spend data: by providing visibility into where and how the team was spending, we were able to bend the cost curve.
Here is the framework we used for successfully cutting cloud spend, what the team learned and how leaders can incorporate data democratization into their cloud spend strategies.
Step 1: Understand the bill
On cloud spend, the first step is to get a clear picture of what you’re paying for through cost allocation tagging. This is easier said than done — tagging one cloud vendor bill is a significant undertaking alone, so multicloud organizations will find this step can be twice or three times more complicated and time consuming. But the output is valuable: This approach — and the data you’re able to garner with it — offers three significant benefits.
Don’t do it flippantly: Reducing cloud spend must be more about optimizing budgets for long-term ROI than short-term cost cuts.
First, it provides visibility into who the owner of each resource, which function it supports and how much the team is using and spending. Second, it enables you to view your spend in different ways, such as organizing the data by service type, cloud, database, network, or department, or viewing month-over-month trends, to understand patterns.
Third, investing time in cost allocation tagging makes it possible to build infrastructure to auto-tag future expenses and easily understand bills on an ongoing basis.
Step 2: Leverage your new field of vision for discounts
Now that you have adequate and reliable visibility into your cloud bills, look into where you can cut and leverage vendors’ native tools to do so.
The goal here is to strike the balance between compute time horizons, high utilization and discount offers. This methodology hinges on economies of scale or ensuring that the team is truly getting the most out of their tool choices.
Next, identify the cheapest option that meets your needs. Whether it’s adjusting timing (scaling up during cheaper timeframes, like weekends) or geography (cloud providers often charge a premium for certain regions), there are several ways to go about this, but there may be hidden challenges based on your roadmap or workflow. For example, you may want to utilize more cost-effective regions for testing, but it’s essential that your product works in the region where you’re going to deploy it — so do this methodically.