This article was published as a part of the Data Science Blogathon.
Cloud Adoption for Data and Analytics
According to a Gartner report, by 2024, 70% of enterprises will use cloud and cloud-based AI infrastructure to operationalize AI, thereby significantly shifting the data gravity to the cloud. Both analytics and BI (ABI) and data science and machine learning (DSML) markets are investing heavily in the cloud today. Incumbent and startup vendors are releasing their latest capabilities cloud-first or cloud only. Cloud ecosystems now dominate spending in the data and analytics space.
Cloud Cost is an Important Concern
This massive adoption has created a lot of opportunities and new governance challenges. On the one hand, it has provided enormous scalability at the fingertips and lowered the entry barrier. It has also moved the organization to consumption-based cloud economics. In this new consumption-based cloud economics, managing costs can be more challenging since purchasing decisions are not governed by a central IT finance or a single procurement team. It is more distributed. The developers or engineers building and running cloud solutions affect costs when writing a line of code or spinning up a new server. This has created a scenario where CFO and the finance team do not have good visibility and predictability on the cloud cost that once they had with centralized procurement teams.
With more and more cloud adoption, this cost is becoming a significant portion of several data and analytics initiatives and can negatively impact the planned ROI.
According to a Gartner report, cloud spending will reach $600B by 2023. Surprisingly, 32% of the Cloud spending is wasted, contributing billions of dollars annually.
A new discipline called FinOps or Cloud financial management is emerging to address this waste. In this function, engineering, finance, technology, and business team collaborate to optimize cloud spending decisions to get the maximum business value from adopting cloud computing. FinOps.org has done great work defining FinOps, its lifecycle, and capability framework.
Implementing FinOps
As I referenced in my earlier articles, FinOps is an emerging area driving much interest from different organizations. Based on State of FinOps reports, 40% of the organizations surveyed plan to grow their team and investment in 2022 and beyond.
Needless to say, this area has attracted a lot of interest from several software product vendors. The FinOps vendor landscape seems to be fragmented, and you have a plethora of options. From matured vendors like CloudHealth, Cloudbility, IBM’s Turbonomic, midsize Flexera, Spot, Snow, Corestack, Cloudwiry, and new startups like Stacklet and CloudThread. Though most provide similar services, a few try to carve out their niche.
Also, All three major cloud providers offer out-of-box tools to deliver cost visibility and optimization recommendations. AWS provides trusted advisors, cost explorer, and compute optimizer, while Azure offers AzureAdvisor. In this situation, as a practitioner, it is natural to wonder whether one should invest in a vendor product or DIY using free tools from cloud providers.
FinOps Technology Choices
Cloud Provider’s Native Tools vs. Vendor Products
The cloud provider native tools from the cloud providers do an excellent job on
- Cost Visibility: With these tools, you can visualize, allocate, and analyze costs across different resources. You can also do trend analysis. Apart from the Azure portal, Azure provides a Power BI app that offers the ability to slice the data based on different criteria.
- Governance: Tools like Trusted advisor and Azure Advisor provide recommendations across all five pillars of the well-architected framework, viz. performance, reliability, operational excellence, and cost. One can configure several policy rules to automate policy violation detection.
- Cost Saving recommendations: These tools do a decent job of detecting some common issues like identifying unused resources, unattached storage volumes, reservation candidates, etc. We can allocate the budget for different resource groups and set alerts to detect any cost anamoly.
- Performance data: Cloud providers give you good visibility of the critical performance matrices like CPU and memory utilization. You can use this data to analyze further right-sizing infrastructure.
- APIs: Additionally, all these cloud providers expose billing and consumption data through APIs. In my current organization, for the FinOps effort, we are exploiting Azure APIs to get more insights into cost management.
However, there are a few gaps in the offerings from these out-of-box tools, and the vendor product can provide specific advantages.
- Actionable and precise recommendations: Most of the time, the cost recommendations from the native cloud tools are not entirely accurate or actionable. Moreover, most of these native tools only analyze up to 14 days of the data. Some software vendors go beyond just analyzing the past 14 days’ data, and they also apply AI/ML models to predict future usage and provide more data-driven optimization decisions. Most of the rightsizing is about scaling the environment down. However, it would be great to have alerts about the resources that need to be scaled up. These software products can do a better job in providing scaling-up recommendations than the native tools.
- Multicloud Management: A lot of organizations are moving to multi-cloud. A lot of these FinOps software vendors provide a single pane of view across multiple cloud providers, and the FinOps team’s job using such products will be much more simplified than trying to use native tools from different cloud providers
- Integrating consumption records beyond cloud providers: Most organizations configure tools like Splunk and DataDog to gain cloud workloads performance, usage, and other telemetry data. Sometimes you would like to marry the data from these tools with the consumption/billing data to answer questions like why the cloud cost shot up on a Monday morning? Some FinOps vendor tools provide a single cloud bills view across cloud providers and data warehouses like a snowflake, observability tools, and other consumption-based platforms. Having one single view of the cost helps immensely to provide complete visibility and drive overall cost optimization efforts.
- Cloud-native support: Many organizations are adopting more cloud-native and container-based architecture. Software vendors like KubeCost and Cast.ai focus on cloud-native and container cost optimizations.
- FinOps at Scale: Leveraging native tools are great. But when rolling out FinOps efforts at scale, these software products and tools can be handy. In most organizations, FinOps is still an evolving function supported by a small team. Moreover, FinOps is not a one-time effort but a continuous process. The FinOps processes should be integrated with your SDLC/DevOps processes. Automation would be a key to making the FinOps program successful by having IaaS, Policy automation, automated anomaly detections, and remediation. Managing reservations is also a critical rate optimization task as it will significantly impact the cloud bill. With thousands of VMs, SQLs, storage, and reservation optimization across them could be daunting tasks for someone in the FinOps team. FinOps Software products can fill in the resources and skill gaps in such situations, enabling you to operate at scale effectively.
Conclusion
Following are the key takeaway.
- Cloud Financial Management aka FinOps is an important focus area for several organizations, and many are increasing their investments in 2022 and beyond
- FinOps efforts can be rolled out with or without investing in the FinOps software product.
- Investing in FinOps software is like the decision to hire a financial advisor to manage your investment portfolio. You can always manage your investments yourself, but professional help gets you expert advice and free you up from everyday efforts.
- You can also start your FinOps efforts with the native tools and go through initial cycles of informing and optimize phases ( as defined in the FinOps lifecycle). For Operate phase or scaling your FinOps practice, you can choose the right software product vendor.
- You have several good choices with the attractive market segment, and you will not go wrong with some due diligence and ROI analysis.
