What Is a Point Release?

Grayscale publishing (also known as canary publishing) refers to a publishing method that enables a smooth transition between black and white. A / B testing can be performed on it, that is, some users continue to use product feature A, and some users begin to use product feature B. If users have no objections to B, then gradually expand the scope and migrate all users to B Come. Grayscale release can ensure the stability of the overall system, and problems can be found and adjusted at the initial grayscale to ensure its impact.

Grayscale release

This entry lacks an overview map . Supplementing related content makes the entry more complete and can be upgraded quickly. Come on!

Grayscale publishing (also known as canary publishing) refers to a publishing method that enables a smooth transition between black and white. A / B testing can be performed on it, that is, some users continue to use product feature A, and some users begin to use product feature B. If users have no objections to B, then gradually expand the scope and migrate all users to B Come. Grayscale release can ensure the stability of the overall system, and problems can be found and adjusted at the initial grayscale to ensure its impact.

Gray period: The period from the beginning to the end of gray distribution is called the gray period.

Chinese name

Grayscale release

Meaning

Grayscale publishing is between black and white

Release method

A way to publish smoothly

Example

A / B testing is a grayscale release method

Get user feedback early, improve product features, improve product quality, allow users to participate in product testing, and strengthen interaction with users to reduce the scope of users affected by product upgrades

1) Define goals

2) Selected strategies: including user scale, release frequency, functional coverage, rollback strategy, operating strategy, new and old system deployment strategy, etc.

3) Screening users: including user characteristics, number of users, common functions of users, user scope, etc.

4) Deployment system: Deploy new system, deploy user behavior analysis system (web analytics), set offload rules, analyze operational data, and fine-tune offload rules

5) Release summary: user behavior analysis report, user questionnaire survey, social media opinion collection, and product function improvement list

6) Product perfection

7) New round of grayscale release or full release

Grayscale publishing and A / B testing commonly used by Internet companies seem to be similar. Foreign Internet companies do not seem to have the concept of so-called grayscale publishing. According to the definition of A / B testing in wikipedia, A / B testing is also called: A / B / N Testing, Multivariate Testing, so the grayscale test can be regarded as a special case of A / B testing. Just to avoid confusion in terminology, talk about the differences between the two.

Grayscale release is the gradual expansion of the range of users for the release of a product, also known as grayscale volume

A / B testing focuses on choosing the best solution among several solutions

For general small systems, there is no need for a separate grayscale publishing engine. You can refer to the practices in A / B testing, and you can implement the rules of flow distribution on the page java script or server side. But for large-scale Internet applications, a separate publishing engine for managing user offloading is necessary.

Partial

1) Problem characteristics:

a The selected sample is not representative;

The b sample is representative, but the user habits of the selected sample do not cover all core functions

2) Solution:

Sample selection should be diversified, and the combination of samples covers most core functions

Curse of knowledge

The "curse of knowledge" comes from the experiments in "Stuck", you can search for it yourself. We are very familiar with the products we develop ourselves, so we take it for granted that users should also be able to understand the design ideas of the products and the functions of the products.

1) Characteristics of the problem:

a results are not quantified;

b relies only on user questionnaires;

c does not have a web analytics system;

d Operational data is not comprehensive, only core business indicators (such as transaction volume), and no user experience indicators

e Analysis of results, choose only information that is beneficial to release, and ignore others

2) Solution:

a product design considers product quantitative indicators

b Results analysis is based on quantitative indicators rather than perception

Post no turning back

1) Characteristics of the problem:

a The user habits of the old and new systems are too different and are not compatible with the original features

b The old and new systems ca nt run in parallel due to the differences in functions, and can only be forced to upgrade

c The new system only implements some of the functions of the old system. Users must use all functions in full and switch between the old and new systems.

d The data structure of the old and new system databases is too different to run in parallel

2) Solution:

Early product planning focuses on these issues, including:

Rollback scheme, compatibility between new and old systems, consistency of user experience, continuity of user habits, compatibility of new and old system data models

Insufficient user engagement

1) Characteristics of the problem:

a Count on users to tap all the features themselves. For a product, most users often use only some functions, and most of them are lazy and do not actively explore product functions.

b single interactive channel

c fall into the "curse of knowledge", disrespecting participating user opinions

2) Solution:

a. Be kind to the sample users who eat crabs, including small rewards for users who participate in the test (such as MS to the Win7 test user genuine license), and title users

Interact with users through new media such as email, forums, communities, blogs, and Twitter

c Provide product function guide. In hotmail's recently upgraded function tip, gmail's tip has similar product function orientation. In the product, a prompt will be similar to: Did you know that xx also provides xx function, through which you can xx.

Gmail Labs is a new feature window. Users can choose some new features that have not been officially released to experience. If they don't like it, they can close it. In the process, they ate crabs and became Google's white mice.

This method is much smarter than traditional grayscale and more respects users:

1. It does not impose a user, whether the user is willing to be a mouse completely voluntary

2. The new feature is not a large version packaged together, you can choose some favorite crabs to try

3. Crabs can be discarded if you don't eat them.

Of course these benefits come at a price:

1. To develop a labs platform to implement new features on the shelves and try the function independently, this may need to change the front and back-end architecture of Gmail

2. New features must be written according to certain specifications before they can be released on this platform, which may increase some workload

3. After the increase of mice users, the pressure on the system may increase, because the interface called by each user is different.

Now that Gmail Labs can be successfully released, it means that these issues are not a problem for Google. In addition, the new features shown now all indicate the developer s name. So, Gmail Labs may open this platform so that external developers can also submit features? This is a very open development mode, which is very suitable for Google's web app product line.

Internet products have a characteristic that they are constantly upgraded, upgraded, and then upgraded: My project team basically maintains the frequency of release once a week. System upgrades are always accompanied by risks, risks of compatibility between new and old versions, and user habits Risk of user loss due to sudden change, risk of system downtime .....

In order to avoid these risks, many products adopt a grayscale release strategy. The main idea is to focus the impact to a point, and then spread it to a surface. It is easy to roll back when an unexpected situation occurs.

For a long time, we have been improving the search engine's ranking algorithm to try to make the best products appear on the first screen of search results. We have tried many algorithms and constantly adjusted the proportion of each sorting factor. But we can't be sure that our ranking results will meet the needs of all users. So we used grayscale publishing to select several first-class product categories and apply different sorting algorithms to them. For example, in the women's clothing category, we adjusted the ratio of seller's credit to 60%, and in the jewelry category , We adjusted the ratio of sales to 60% .. Then published it, collected user feedback, and finally chose an algorithm that most people think is good.

QZone is another example of grayscale publishing. Everyone knows that QZone has improved tremendously in the past year

Grayscale release (2 photos)

Yes, from the old slow grandpa to a young man full of vitality. Among them, there have been numerous releases, and their releases have also adopted a grayscale release strategy. The upgrade of user data is not a large-scale one-time upgrade, but a user upgrades the logo server. If the user data is not upgraded, , The background will gradually migrate this user's data to the new version, and then set the upgrade flag to 1, during the upgrade process, users can still access the old data, access after the upgrade is completed will be forwarded to the new version.

Many of QQ's product releases use grayscale releases, some are extracting some QQ number segments to upgrade to the new system, and then upgrade in a wide range according to user feedback.

In the traditional software product release process (such as the release process of Microsoft Windows 7), it generally goes through several stages such as Pre-Alpha, Alpha, Beta, Release candidate (RC), RTM, General availability or General Acceptance (GA), etc. (Refer to Software release life cycle). It can be seen that the release phase of traditional software is from the company's internal-> external small-scale test> external large-scale test-> official release, the number of users involved is also a process of gradually increasing volume. This method of release is also used in the process of Internet product release: the product release process is not done overnight, but gradually expands the scope of users, from internal users of the company-> seed users with higher loyalty-> larger scope Active users-> all users. In this process, the product team improves product-related functions in a timely manner based on user feedback. According to the Chinese characteristics, this method of publishing is titled "Grayscale Release", "Grayscale Volume", "Diversion Release".

There is no way to investigate the origin of the term "grayscale release". However, according to the traditional Chinese philosophy, it is in line with the Chinese moderation mode of thinking: all things in nature always exist in symmetrical, complementary, and harmonious forms, such as black and white, yin and yang, positive and negative, and happiness. And woe. In the stage of mutual transition between the elements of binary opposition, the so-called "Blessings of Blessings, Blessings of Blessings". Specific to black and white, there is a neutral color in the middle of black and white-gray. So there have been many sayings about gray: gray box testing, gray management (recommended Ren Zhengfei: gray management), gray income, gray area, and so on. Therefore, grayscale release is actually a process of gradually transitioning from never-to-publish to official release.