Technology cloud report original.
On the forum, a real case of a foreign company sparked heated discussions. At the beginning, the company only cared about technological innovation and actively migrated to the cloud, regardless of cost.
Until one day, high-level executives intervened and stopped: "This cloud can no longer be accessed, and the cost has far outweighed the benefit." The company's cloud migration progress was delayed by two years due to cost out of control, which seriously affected the company's technological innovation.
As cloud computing becomes more and more popular among enterprises, enterprises will find that the cost of using the cloud is also rising, which seems to run counter to the concept of "reducing IT costs" that was announced when cloud computing was born.
Because of this, how to effectively control cost, quality, and efficiency in the cloud era has become a new topic for enterprises to use cloud to manage cloud. The corresponding cloud cost optimization (FinOps) term is also becoming more and more popular.
On Google Trends, the search volume for the keyword “FinOps” has increased by 410 times in the four years from 2019 to 2023. In foreign countries, more than 18,000 people have listed FinOps skills in their LinkedIn resumes.
In the cloud native 2023 trend forecast report released by CNCF, 4 of the 10 hot trends are related to FinOps, namely FinOps, GreenOps, GitOps and cost reduction.
Let’s talk today, what is FinOps, and how should companies practice FinOps?
FinOps: Create maximum value at minimum cost
The history of FinOps is not long. Adobe and Intuit, early users of public cloud, first sketched the prototype of FinOps in 2012. FinOps is essentially a theoretical framework without a specific technology stack. Its methodology comes from the integration and abstraction of the best practices of various cloud vendors, and it provides suggestions in terms of organizational processes, waste identification, and optimization measures.
FinOps defines a set of cloud financial management rules and best practices that enable organizations to maximize benefits by enabling engineering and finance teams, technology and business teams to collaborate with each other to make data-driven cost decisions.
This diagram from the FinOps Foundation has been quoted many times, and it simply lists the principles, goals and participants of the FinOps theory.
Image source: FinOps Foundation (Chinese translation)
The ultimate goal of FinOps theory is to create the greatest value at the lowest cost, and points out three stages of cost optimization:
●Cost-aware nodes focus on cost visualization, cost allocation, etc.;
●The stage of cost optimization can focus on target formulation, and then save costs through rate optimization and usage optimization;
●In the operation and maintenance phase, continuous cost optimization is achieved through continuous optimization of processes, norms, and resource operation methods.
At the same time, FinOps theory also has some maturity assessment models to assess whether the enterprise is doing well.
These three aspects are widely involved and difficult to implement. It is a systematic project that requires the participation of all employees of the enterprise. Therefore, the prerequisite for success is the high alignment of organizational goals, the establishment of business awareness among all employees, the organization's firm execution and continuous improvement of execution. Efficiency, practice itself is a great training for organizational efficiency.
How is FinOps implemented?
Flexera, a well-known IT software company, has conducted an annual survey of cloud computing decision makers for 12 years. According to the "Flexera 2023 Cloud Computing Status Report" released on March 8, 82% of the respondents said that the biggest challenge they face is is managing cloud spending.
Nearly half (45%) of respondents said they expect cloud usage and spending to be slightly or significantly higher than originally planned due to economic uncertainty.
So perhaps it's no surprise that cloud cost management is getting a lot of attention. Today's economic volatility means that while cloud usage and spending remain robust, businesses are increasingly sensitive to the costs associated with it.
So, when it comes to cloud cost optimization, how to reduce costs and increase efficiency?
There are two ways for enterprises to reduce costs and increase efficiency: one is to reduce production and reduce waste. Foreign survey reports show that at least 35% of cloud resources are wasted. The second is to reduce prices, starting from the billing model to optimize.
After establishing the path, how should the enterprise implement it? Some professionals in the industry have given the following steps:
First, all mobilization . Include this role or organization or team that should be involved.
Second, build an accurate panoramic map of IT resources . Construct a global resource map of the enterprise through CMDB to facilitate communication between various teams, or to ensure that everyone's information is aligned when talking about cost reduction and efficiency increase of a certain project or a certain environment.
Third, reasonable labels . The allocation of costs is based on the system's hierarchical architecture, technical architecture, and business architecture on the panoramic map of IT resources. Among them, a set of labeling systems is very important, and it needs to be regarded as an important daily work. .
Fourth, effective IT resource utilization monitoring . In many cases, enterprises focus on the observability of IT resources, and people pay more attention to system availability monitoring or performance monitoring, and do not pay much attention to the monitoring of resource utilization. If even utilization monitoring is not accurate, then it is impossible to judge which resources are wasted.
It is worth noting that utilization monitoring for FinOps is different from traditional operation and maintenance monitoring. Traditional operation and maintenance monitoring pays more attention to the average utilization rate, while FinOps pays more attention to the peak value. If the capacity is not calculated according to the peak value, the system will likely collapse after reducing costs and increasing efficiency.
In addition to the optimization method of reducing volume and price, there are some neglected "money-saving ways" that also need to be paid attention to by enterprises, such as:
● Choose a multi-cloud architecture that suits you . Not all businesses are suitable for going to the cloud, and some businesses may be more expensive after going to the cloud. At the same time, it is necessary to pay attention to the greatest common divisor of multi-cloud to ensure that it can be cross-cloud without being locked by a certain cloud vendor.
● Make good use of hosting services . For example, after some enterprise data centers go to the cloud, they entrust the machines in the original data center to the service provider, and go to the cloud based on the original machines, which can save a lot of cost.
FinOps practice of well-known Internet companies
Although FinOps is not mentioned much in China, as early as December 2020, the China Academy of Information and Communications Technology took the lead in establishing the FinOps industry promotion phalanx to promote large-scale practice.
In those big Internet companies that took the lead in embracing cloud native, the seeds of cloud cost optimization have actually taken root and sprouted long ago, forming a best practice methodology. The emergence of FinOps has enabled the optimization experience of major manufacturers to be expressed more systematically.
Taking ByteDance as an example, they already have relevant internal practices, such as cloud bill analysis, review of pricing strategies of different vendors under a multi-cloud architecture, mixed deployment of offline tasks for recommendation, advertising, and search, etc. Currently, ByteDance's best practice in cloud cost optimization will provide external services through the volcano engine.
Ali Group has also built its own hybrid cloud resource management platform (HCRM) to promote the construction of its own cost digitization from scratch, and to unblock the cloud resource billing and settlement links within the group.
Within Tencent, the cloud business cost center assumes the responsibility of the FinOps team. It needs to carry the assessment indicators for resource optimization, starting from the platform side and business side, and can even report upwards and promote it through the GM level.
Taking Tencent as an example, it has built a rich cost and utilization rate performance dashboard internally, and shows the performance every day, and will promptly disclose whether it is doing well or not.
Tencent's internal cost dashboard mainly includes two dimensions: the first is which account bought which resources, and the second is which businesses use these resources, including some allocation details.
In addition, there are platform and business-oriented maturity indicator boards such as utilization rate and maturity, mainly to understand the overall situation of the resource market, to see whether the part put into use is well used, and to revitalize idle resources and reduce waste.
The FinOps capabilities provided by the platform side help businesses and platforms achieve their goals from the following perspectives:
Business optimization : Provides a special page for resource optimization on the cloud console, predicts based on the history of business resource usage, builds business resource portraits, and gives resource optimization suggestions.
Specification suggestion : By comparing the application volume and usage of business resources, the business can be told the cost data that can be saved, and then the business can be directly optimized through the system console.
Elasticity suggestion : For example, resource usage is very high on a certain weekday, but there is basically no traffic on weekends. At this time, the capacity needs to be scaled down on weekends. These businesses can also be optimized through the console.
Platform optimization : The cloud platform provides many scheduling capabilities based on resource portraits when performing business scheduling.
Scheduling optimization : A dynamic scheduling capability oriented to the real utilization rate is proposed. The administrator sets the target utilization rate of nodes. As long as the utilization rate has not reached the target, the scheduler can schedule more services to come in.
Mixed department capability : Introduce differentiated SLA, allowing high-quality online business and low-quality near-offline business to be mixed to squeeze out every bit of computing power. At the same time, offline services can immediately relinquish resource requirements when resource competition occurs, and realize zero online business. interference.
It is reported that Tencent's internal online business uses scheduling optimization methods to pull the resource utilization rate to 48%. Coupled with offline hybrid deployment, the resource utilization rate of some clusters can reach more than 65%.
Overall, Tencent's CPU scale has reached 50 million cores, and cloud cost optimization has saved a total of 3 billion yuan.
When will enterprises end FinOps?
Although the big Internet companies have set an example and achieved considerable results in the practice of FinOps, does this mean that all enterprises should enter the practice of FinOps now?
In fact, the focus of enterprises on cloud migration is gradual: in the first stage, enterprises pay more attention to privacy, stability, and industry regulatory policies; Focus on the improvement of efficiency; the third stage will come to the level of cost optimization.
At present, most domestic enterprises are still in the period of migrating their business to cloud-native. When enterprises face business pressure, the priority of cost control will naturally be “released” first; The modernization of the company is nearing completion, which is when it is time to turn the focus to cost optimization.
But the status quo does not mean that this is the ideal state.
Some industry experts bluntly say that cloud cost optimization should start planning from the first day of going to the cloud and continue to optimize.
"In the process of migrating to the cloud, many companies simply apply past experience to the new technology stack. There are also many teams who arbitrarily believe that cost optimization and business stability are in conflict, and they can only choose one of them." The expert said.
epilogue
FinOps is here to stay and is in the early stages of rapid development. For enterprises, early practice and transformation will always bring pains, and practitioners need to be mentally prepared for this. How FinOps can help enterprises make good use of the cloud in the future still requires the unremitting efforts and exploration of the entire industry.
[About Science and Technology Cloud Report]
Focus on original enterprise-level content experts - technology cloud reports. Founded in 2015, it is the top 10 media in the cutting-edge enterprise IT field. Recognized by the Ministry of Industry and Information Technology, Trusted Cloud, one of the official media designated by the Global Cloud Computing Conference. In-depth original reports on cloud computing, big data, artificial intelligence, blockchain and other fields.