12 tips to avoid cloud traffic congestion

In the story of cloud computing, the process is quite smooth, you only need to gently press the button. This is true for simple applications that are independent and do not rely on external systems or third-party application interactions. Even for some of the more complex cloud environment deployments, you can also follow the following points to avoid traffic congestion:

1, check the relevant qualification certificate. To avoid the deployment of traffic jams and accidents, the qualification audits of the relevant deployment vendors are important (even in most cases it is necessary to check the developer's Certificate of Competency).

2, involve your development team in the architecture review of proposed changes. Some seemingly innocuous changes, such as access control, validation rules, and even the choice of list values, can have a significant impact. If your organization has more than one development team working in parallel, the impact will be greater.

3, if you can find a configuration management tool (or at least an audit tool) in your cloud computing technology environment, use it as much as possible. If you can't use it, use a version management file to track all changes. Docs is pretty good, especially if your profile is a static file and needs to be archived weekly.

4, Unit testing of all modules, more than 90% test coverage. Assert testing of major positive and negative test cases. For core business logic, you can consider using test-driven development.

5, all unit tests run every day, and the results are recorded, even if your code does not change. Because your SaaS platform or Third-party software element may have changed, and you don't know it. If any unit tests fail, the development team must rank them as their highest priority processing event.

6, periodically deploy the system with ant or some similar scripts, and try to follow the best time scenario for continuous integration (CI) agility.

7, system-level testing of each external interface with representative test data, and performing database statistics checks (for example, "Do we have an integrity check for 50% of new customers since the last run performed?").

8, run tests on all systems at least once a week and record the results, even if your code does not change. Since not only your platform or Third-party software has changed, your system's data may also have a corresponding expansion or change in a previously untested way of calling code paths. If any unit tests fail, the development team must rank them as their highest priority processing event.

9, always keep synchronized/refreshed. The synchronization/refresh cycle should be explicitly incorporated into the project plan and become a development teamPart of the team's agenda.

10, a systematic management configuration control measures to prevent experimental changes, adhere to beyond the experiment.

11, effective control to prevent illegal alteration of production systems.

12, effective control prevents changes from being directly applied to production systems.

What to do in the real world

Cloud vendors are as innovative as they are, and they are still in the early stages of deploying infrastructure and rules. The best cloud providers should be doing their job for their direct customers, but there are not many vendors with good technology, let alone open Source services.

Too many enterprises simply do not have the right to deploy the infrastructure and the points listed above are in place. If this is the case, you need to assess the risks your business is currently in and which of the corresponding points should be corrected first.

If your team is already in trouble (for example, you cannot deploy a bug fix), the first priority is to get out of the trap of being bothered by this local problem alone. In this dilemma, the rule you follow is to focus on the control of the situation and actually not get where you want to go. You need to figure out how to simplify the problem, reverse the situation, and reduce the number of variables. Once you have solved this problem, be sure to prioritize the deployment of infrastructure and practices prior to developing any new features. And why? You have built a debt that technology must pay. Ensure that each new project pays the "deployment cost", so that the deployment infrastructure and related rules do not suffer from resource scarcity and can enhance the challenge in the relevant moving parts. Otherwise, you will never get out of the deadlock. The

David Teber is the author of the new publisher's book, "Salesforce.com's Secret of success," and is the chief executive of sales Logistix. Sales Logistix is a consultant for Salesforce.com website certification, and has long been concerned with promoting business processes through the use of CRM systems. Sales Logistix customers throughout North America, Europe, Israel and India. David himself has more than 25 years of High-tech industry experience, of which 10 years in the VP and above grade.
 

(editor: Heritage)