Fresh off the quarterly earnings for Microsoft, Google and soon Amazon — where cloud spend is slowing as companies focus on cost optimization — we are releasing the Q2 2023 Cloud Cost Report, an analysis into cloud usage based on anonymized Vantage customer usage. Vantage is a cloud cost management and optimization platform with a unique view into industry trends thanks to tens of thousands of connected infrastructure accounts across 11 cloud providers. To discuss this report in more detail, join our growing Slack Community of over 1,000 engineering leaders, FinOps professionals, and CFOs. View past reports here.
Get a PDF of this report.
It's never been easier to choose the right cloud for the right workload. In Q2 we measured the top services across AWS, Google Cloud, and Azure and found a different mix of services dominating share of costs on each.
"With Google Cloud, we’ve been really embracing an open architecture. We’ve embraced customers wanting to be multi-cloud when it makes sense for them."
- Sundar Pichai, CEO at Google during their latest earnings call
With Google Cloud, BigQuery helps drive big data workloads that then also consume spend for Bigtable, Spanner, and Cloud SQL. On AWS, the ranking of spend is more traditional with compute, data, storage, and monitoring services dominating the top 10. Compute, whether EC2, Compute Engine, or Virtual Machines consistently draws around 30% of spend. Organizations are increasingly becoming more multi-cloud focused, choosing certain clouds for certain portions of their workloads.Tweet
Datadog is known for its comprehensive toolset and high and complicated pricing. Similar to traditional IaaS providers, the high price is driven in part by customers who are running too much of their Datadog services on demand. This chart shows the share of committed vs. on-demand spend for popular Datadog services for which there is a committed-use option available.
Only "Infrastructure Monitoring" has more than 50% of its spend covered by Committed Use Discounts. "Containers" and "Synthetic Monitoring" in theory should serve static workloads (see Datadog cost optimization tips) so customers could seek out higher rates of coverage. To take advantage of these discounts, talk to your Datadog rep.Tweet
The last year has been a chaotic time where organizations have been performing optimization en masse. Recently, spot instance prices have trended higher to converge with on-demand rates. After customers overwhelmingly made investments in committed usage, we're beginning to see on-demand spend increment back up over 40%
On average over the past year, a little under 40% of the spend we recorded on EC2 came from on demand instances. This may be the result of optimization fatigue or the impact of the spot market pushing customers back to on-demand.Tweet
In previous reports we have tracked the slow wind-down of m5, c5, and r5 instances in favor of newer generation ones. At first, only ARM-based Graviton instances were available among newer instance classes but with the introduction of Intel and AMD instances added to the mix, customers are now rapidly upgrading their workloads.
c6 and m6 instances have each hit almost 20% of costs in the graph which tracks the share of EC2 costs contributed by each instance type for the first 6 months of 2023. As newer generation types consistently mean better price-to-performance rations, its no wonder why organizations are taking advantage of this more-so these days.
While we have previously tracked Graviton adoption for EC2, where the vast majority of spend for that service still goes to Intel and AMD based instances, there is a different story for RDS, ElastiCache, and OpenSearch.
In fact, we found that an astounding 77% of OpenSearch costs and 57% of ElastiCache costs came from Graviton based instances in Q2. Looking deeper, we can see a few reasons. For one, there are no Intel or AMD instance types available in the newest generation for these services. Secondly, this graph reflects the degree to which certain services can be upgraded. For example, it's probably easier to upgrade a cache vs updating your RDS database to capture cost savings.Tweet
With RDS as the 2nd most costly service on the Cloud Cost Leaderboard we thought it made sense to peek into those costs and see what was driving them.
The key to cost management on RDS is to balance the size of instance with the required IOPS and storage for your workload. If this balance is off you can end up with overprovisioned IOPS or oversized instances. Aurora promises to solve some of these issues but adoption is still nascent in our cost data. Recently, AWS went a step further and released Aurora I/O Optimized Instances which may finally be the perfect combination RDS users have been looking for.Tweet
Since the release of the GP3 volume type for RDS we have been tracking its adoption and this quarter we saw a doubling of the share of RDS storage costs on GP3 with a continued decline in GP2.
This is great news as the introduction of new volume types can sometimes take a while to become the standard practice. These volume types offer better performance in certain situations or at least on par performance from GP2. It's most likely worth it to upgrade.Tweet
Get the upcoming 2023 Q3 Cloud Cost Report directly in your inbox.