Get rid of technical thinking and turn to product thinking--looking for the distress of "omnipotent" IDC

Background: The recent development of new products after the mission has been looking for good IDC and high-quality hosting services busy. The demand comes from our focus to solve the problem that the old version of the product faced by users at home and abroad is slow or not connected.

In addition to the use of high-performance, scalable scenarios for the architecture technology, a better user experience can be achieved only by ensuring high quality of the network quality for the current concurrency and data volumes.

First of all, our product platform data need to be strong consistency, high availability. And very much is dynamic data.

Based on the CAP theory, we first vetoed the distributed deployment scenario, one of which is the lack of self and team's grip on distributed technology, and the second is that we disagree with the research and design that we have too complex from the development cycle. Therefore, the goal of moving to a strong consistency of data and services to adopt a centralized deployment of the scheme, so in the face of domestic network interconnection problems, even in the face of international network interoperability problem we have to rely on third-party services-good IDC and hosting services. In the experience of traditional solutions in the multi-line room. The test certificate of the BGP room (which has been suspected not to be a true BGP room). Also experienced a domestic well-known manufacturers of CDN Acceleration Technology test Certificate, found not reliable, connection speed and connection is not the problem even our own small-scale test will meet.

Melancholy in a careful study of the following various managed services indicators, listen to some of the professionals answer suggestions, their hands calculate the next network quality, think of the company to the hosting services this piece of the expense of tolerance (to know that the BGP room is very expensive, CDN acceleration service Price is more frightening, more scary is the protection of the end).

First of all, the BGP room, from the contact to the BGP room view are the optimization of domestic lines support. Even if we can deploy the service to the backbone of the country and country, our country's export bandwidth and the terrible wall can not say it will affect the quality of the network, plus more than 30,000 kilometers of submarine cable transmission (for example, to the United States), the forwarding of each node, ideally, how to delay also get hundreds of milliseconds. Moreover, we dare not guarantee the network quality and service of the end-users abroad. Inevitably there will be a foreign user connection slow or not connected to let users hate, let the bad situation of leadership. As for CDN acceleration Services. According to their technicians, the acceleration of the TCP layer has been customized, and has not been very secure.

In the further melancholy I was thinking of a question: "We can not make products into the domestic and international version, the two parts of the data and services to separate, the nearest deployment server can not solve the problem first." I think we've seen a lot of similar product designs. For example Evernote and impression notes, online games of different areas of clothing .... Then I hastened to the previous product personnel to understand the use of the previous product background, the original they have to worry about doing this after the situation will encounter this: for example, an American in the United States with the international version can have a very good experience, such as fast login, access to a variety of information fast (of course, The server is in their American high-quality engine room). As a result, the Americans came to China one day, because the product landing slow or not connected to our products will have a bad impact (in fact, I want to China's Americans know that there is a VPN this thing).

I'm not quite able to agree with this product concept. Can not be a product design to meet a variety of extreme conditions of the perfect product, we think of our own visit to the foreign server on the things slow to know, you can understand such a situation? Later assumptions such a situation is very many, indicating that our user volume also came up, but also the money we think of other ways.

In addition, for the product design, I think it is also worth to be divided into domestic and international version. We are now the design of the people all over the world with the same function of the same client (of course, different languages), there is a saying that "product without national boundaries, but users have national boundaries", now our product design does not have a specific positioning of users, often think of adding a feature all users have to upgrade. Domestic users are okay to say, or we have different life background, cultural background of foreign users?

Something: Now we're still testing the "universal" hosting service. I have also conveyed my ideas to the relevant people. Got part of the affirmation but also some disagree, in short I think in the development of the hypothesis can rely on product design and business adjustment to avoid some complex technical situation is worthwhile.

Get rid of technical thinking and turn to product thinking--looking for the distress of "omnipotent" IDC