What is Grayscale publishing?

Grayscale Publishing is a way of releasing a smooth transition between black and white. AB test is a grayscale distribution, so that a user continues to use a, some users start with B, if the user has no objections to B, then gradually expand the scope of all users are migrated to B above. Gray-scale release can ensure the stability of the whole system, in the initial gray level can be found, adjust the problem to ensure its impact.

Gmail Labs is a new feature window, users can choose some http://www.aliyun.com/zixun/aggregation/18782.html "> officially released new features to experience, do not like to be closed, In the process, eating crabs, but also when the Google mouse.

This approach is much smarter than the traditional grayscale and more respectful of users:

1, it does not strong X users, whether the user is willing to be a mouse completely voluntary

2, the new features are not packaged together in a large version, you can choose some like crabs to taste

3, crabs are not good to throw away, do not eat into the stomach caused by gastrointestinal inflammation

Of course, these benefits also have a price:

1, to develop a labs platform to achieve new features shelves, independent attempts to function, which may have to change the front and back of Gmail architecture

2, the new features to be written in accordance with certain specifications in order to publish to this platform, may increase some workload

3, mice after the increase in user, the pressure on the system may have a certain increase, because no user calls the interface is not the same

Since Gmail Labs can be released smoothly, it is not a problem for Google. In addition, the new features that are now displayed are the names of the developers, so Gmail Labs may open up the platform for external developers to submit features as well. This is a very open development model, very suitable for Google's Web App product line.

Internet products have a feature that is constantly upgraded, upgraded, and then upgraded. 　　My project team, basically maintain a weekly release frequency, system upgrades are always accompanied by risks, the new and old version of the compatibility of risks, users use habits suddenly change the risk of user loss, the system down machine risk .... In order to avoid these risks, many products have adopted a gray-level strategy, the main idea is to focus the impact on a point, and then spread to a surface, the unexpected situation is easy to fall back.

For a long time, we have been improving the search engine sorting algorithm, so that the best products appear in the search results of the first screen. We have tried a lot of algorithms to constantly adjust the proportion of each sort factor. But we are not sure that our sorting results will meet the needs of all users. So we use a grayscale release, select a number of commodity categories, in which the use of different sorting algorithms, such as in women's category, we put the proportion of sellers credit to 60%, in the jewelry category, we have the proportion of sales to adjust to 60%. Then publish, collect user feedback, and finally choose an algorithm that most people think is good.

Qzone is another example of grayscale publishing. As we all know, qzone in the past year to improve is huge, from the previous leisurely grandfather into a youthful vigor of the young man. Which went through countless sizes of releases, their release is also the use of grayscale publishing strategy, the user data upgrade is not a large area of a one-time upgrade, but through a user upgrade flag server, if the user data is not upgraded, the background will be the user's data gradually migrated to the new version, and then upgrade the logo location 1 , the user can still access the old data during the upgrade process, and the access to the upgrade will be forwarded to the new version.

Many of the QQ products are published in grayscale, and some are part of the QQ section to upgrade into a new system, and then a wide range of user feedback to upgrade. Most of our products are also used in grayscale release.

This article is from: http://enki-ding-yeah-net.iteye.com/blog/1114565