Gray-publishing needs

What is gray Published

  • What is gray publish it? To understand this problem we must first understand what is gray. Gradation is understood that a smooth transition exists in the area between black and white from the literal meaning, so that the products, for the Internet, on-line and on-line is not black and white points, not on the line to achieve a smooth transition function way is called the gray release.
  • Black and white is never a universal phenomenon, from the perspective of color, grayscale, black refers to an unsaturated, we have a black color as a reference, each object is a gradation from white (0%) to black (100%) intermediate value, which is the middle 98% of ash.

Internet product of the following characteristics:

  • Large-scale users, version update frequently. The new on-line version of each product to be under great pressure, and gray publish good avoid this risk.

In the understanding of what is the definition of gray released later, you can find out how to do this since the release of the gray. Can be drawn through many forms part of the user, for example, to choose their own VIP user, or select a number of active users, these users into two groups, which put A number of versions, put in another group B version, it is necessary prior to delivery a variety of data collection records that may exist to do the work, in order to see both versions of user feedback after launch data to determine which version to use the last to be served by a lot of data analysis and investigation.

What is gray publish it?

  • In general, a complete gray release is required to carry out the necessary user identification, ie distinguish between users, for example, to distinguish from the many ways to pay or the number of your area as well as the level of activity and so on, this distinction also head to to be more precise analysis of the necessary data.

A complete set of gray-release mechanism will include the following stages:

  • User ID : mainly distinguish between users, as well as auxiliary data analysis.
  • Target user / flow Filter : need to refer to user characteristics, user traffic, user scope and consistency of user experience, the iterative version for all users or part of the user, a small flow through the test again heavy volume, in general, according to internal users - seed users - active users - All users of the order is a typical range control and experience consistency asked to consider whether the old and new versions of the span is too large, the user can accept.
  • Real-time data monitoring : monitor the stability of the new version, server stability, frequency of use, frequency of use, such as a comparison of the data with the original data.
  • One-click publishing / rollback : deciding whether to release / rollback data feedback from the results.

Some have questioned the gray release is a waste. But this is not so much a waste as it is redundant and resilient, able to release the gray line to avoid the risk of a new version of the full amount of flow through the small verify the way, can be found in the gray phase, adjust and optimize the product in question, smooth iteration.
While also working to collect all the relevant data, such as a new version of stability, stability and frequency of use of the server, using the frequency and variety of data, convenient and previous original data for comparison.

Some people may think that the gray release completely unnecessary, a waste of resources, where the gray release is useful to do so head to not only understand the most realistic user experience but also can effectively prevent a significant impact on the system BUG more retreated or cause other unnecessary economic losses, so that the gray release is an effective way to avoid the risk of a new version of the line, can first be tested by a small flow of work, to help complete the new version of the smooth iteration.

Guess you like

Origin www.cnblogs.com/linbo3168/p/11929304.html