With the development of digitization and informatization, R&D efficiency and cost reduction and efficiency increase have increasingly become the focus of enterprise management. Especially for R&D teams, delivering value quickly, with quality and quantity is the highest priority task, but in the actual development process, we always encounter situations such as technical debt and parallel conflicts that affect R&D efficiency.
Today, when we bid farewell to barbaric growth and advocate intensive cultivation, how should enterprises/organizations interpret various performance barriers and formulate replicable solutions? This article will start from 7 common R&D scenarios and share the experience and experience on improving R&D efficiency.
Scenario 1: Parallel development leads to code conflicts
Intra-group/inter-group parallelism, or various parallel development caused by code rollback/merge, etc., leading to code conflicts is one of the common performance problems. Parallel branch management and version management are relatively important issues, and merge strategy, feature branch management, and change management may all affect R&D efficiency.
To solve this problem, the following three optimization methods can be considered:
1. Timing serial management
Taking the time as the axis, the entire version main line is strung together, the code is responsible for the version, and the version is responsible for the function.
For the same system, the code is developed in parallel, but the final deliverables/releases are released sequentially; for different systems, the main consideration is the interdependence, the impact and the release order.
2. The function is divided into zero
Break down large functions into parts in an agile iterative manner to better cope with changes. If the requirements must be changed during the iteration cycle, the impact scope of the change and the priority of the requirements need to be determined.
3. Demand divide and conquer
Technical/optimization requirements and follow-up iteration requirements may require different release strategies and branch management. This can not only ensure that business goals are achieved on time and effectively, but also ensure that various optimization and support work can be performed flexibly and concurrently.
Scenario 2: Technical Debt and Architecture Corruption
Technical debt is a clichéd topic. In the normal R&D management, enterprises should pay attention to the cultivation of "good habits". If the technical debt piles up and the system is terminally ill before starting to solve it, I am afraid it will be too late.
It is suggested that in the daily R&D management, the code review mechanism should be strengthened, and the P3C standardization inspection of the code should be implemented; the technical solution of the business should also be reasonably selected in the early stage.
In addition, the architecture design should be fully considered in combination with actual business and resources, and beware of over-design. Good architecture is evolutionary, there is no perfect architecture once and for all.
Scenario 3: Frequent Troubleshooting Tasks
When collaborating in parallel, asynchrony of configuration and resource files is also an important factor causing conflicts and problems. To avoid additional exclusion work affecting R&D efficiency, companies can consider improving the independence and simplification of configuration and resources.
First, try to manage the unique values of the required configuration in chronological order; if the unique configuration cannot be guaranteed, it is recommended to manage the modified values of each group according to the grouping logic (the original configuration of other groups is not redundant).
For example, it can be managed by time series or grouped and managed side by side, and then merged after the detection node is determined. In this way, conflicting items can be found more clearly and prevent each other from being overwritten.
Also, in addition to common configuration, consider grouping configuration by function and don't write everything in one configuration file.
Second, when configuring merging, the check-in and check-out process should be as short as possible. The configuration merging process needs to be reviewed, but the time window for the configuration adjustment process should not be too long, so as not to cause additional waiting costs and induce potential conflicts.
Scenario 4: Troubleshooting Production Problems and Data Security
In many cases, the data in the production environment must be desensitized, but at the same time, the R&D team needs to verify the production problem or reduce the impact of the problem. How should this situation be interpreted and resolved?
1. Complete verification with desensitized non-sensitive data
After the customer data in the production environment is desensitized, some insensitive ID parameters, exceptions, and other logs can still be recorded as valid data to complete analysis requirements in specific scenarios.
2. Synchronize non-customer data in Pre production environment
Prepare a "clone" that is relatively consistent with the production environment - Pre quasi-production environment, synchronize and complete the verification of the production environment through non-customer data.
Non-customer data includes part of production test data, part of data that can be collected with the permission of the customer, and data that has been completely desensitized for compliance, etc.
3. Use A/B testing to penetrate and run first
Through the penetration of a small number of customers, or a short-term penetration operation of the production environment for some specific tenants, and then put into large-scale deployment after stabilization.
Scenario 5: Environment Complexity
During the R&D process, the complexity of the development environment and deployment environment will also affect the R&D efficiency. Therefore, it is recommended to minimize the complexity of self-testing, joint debugging, and environment deployment, as well as the code size and complexity of the same service.
For example, some systems take 30 minutes to start up, so the time cost each developer spends per day in dealing with the environment and dealing with the startup also increases significantly.
Scenario 6: Production Issues and Potential Issues
It is undeniable that there is no product or service that will never fail. Therefore, it is very important to build an effective and quick-response business monitoring and operation and maintenance monitoring system .
No matter which monitoring platform system is selected, the core purpose is to monitor the core target and realize timely early warning and notification of key indicators. Effective, direct, and rapid response and handling of discovered problems is more important than a rich monitoring program.
Second, focus on testing. Consider adding a variety of testing methods to find problems as much as possible, such as automated testing for interfaces, integration testing for scenarios, and stress testing environments for large systems.
Scenario 7: Non-technical influencing factors
In the process of R&D process management, non-technical factors will also have an important impact on R&D efficiency.
- The simplicity and rationality of the R&D process
- Continuous product output and reasonable demand granularity
- Cost and Loss of Meeting Efficiency and Communication and Coordination
- Differences in effective communication styles caused by different personalities
- chronic tension
LigaSummary
R&D process management is an important topic in the field of R&D efficiency improvement. Managers can use a bird's-eye view to analyze and judge the operation of the entire R&D life cycle, and with the help of intelligent monitoring and early warning tools, they can find, solve and avoid problems, and make more reliable management intervention and guidance.
To learn more about agile development, project management, industry trends, etc., follow us at LigaAI@oschina or click LigaAI - a new generation of intelligent R&D collaboration platform to apply online to experience our products.