| The Cloud computing Road series blogs share the actual experience of moving the site from IDC to the cloud computing platform, and this time we share the reason why we choose cloud computing.
This blog post shares five considerations.
first, effectively solve the hardware single point of failure
A single point of failure is the failure of a particular hardware to cause a service interruption in a Web site. To really solve this problem, you need to prepare redundancy for each hardware, which not only increases the cost of hardware acquisition, but also allows for deployment and maintenance costs.
The cloud computing platform is based on the server cluster, from the beginning of the design to consider a single point of failure, and in the construction of the effective solution to this problem. If a cloud service provider had a single point of failure, it would be like having a bank lost money.
Ii. Increase/decrease of hardware resources on demand
Hosting your own server, adding/reducing hardware has been a headache.
1. When the server is added, it takes time to buy the server and it cannot be controlled at this time. Using a cloud server, you can add servers at any time--at your fingertips.
2. Reduce the server can only be pulled back from the computer room to the office, can not return the server to the manufacturer, the cost of the purchase of the server is wasted. And the use of cloud server, if not next month, do not renew the line (for Aliyun monthly purchase of the situation)-to use it, want to throw it.
3. The server configuration that meets the basic requirements cannot be increased on demand. If we need a low configured server to run caching with Linux, it would be uneconomical to buy a cheap, low-configured server for it, because the server will cost at least 3000 yuan a year. So you can only minimize the number of servers, improve the configuration of a single server, in let a server run more things. And the use of cloud server, what kind of configuration to buy what kind of configuration, so that the responsibility of each server is more unitary, the impact of each other smaller--the responsibility is clear, efficiency is higher.
third, the BGP route solves the north-South interworking problem
North-South interoperability is the South Telecom and the North China Unicom line between the exchange problem, this problem has plagued us for many years, before using a dual-line room, the solution is not very good. At present only the BGP line can effectively solve this problem, and the real BGP line is not a lot of rooms, the cost is very high. And the Aliyun I'm going to use is the BGP line, which is one of the main attractions for us.
What exactly is the question of North-South interoperability. Based on our understanding of simplified description, the wrong place is welcome to point out. North-South interoperability is actually a routing problem. Suppose our server is in Shanghai Telecom's room, a Shanghai Unicom users visit our server, we must first around the Unicom's Beijing total exports (assuming total exports in Beijing), and then back to Shanghai. In fact, the Unicom users can get through the Shanghai line directly to our server, not around so far, but Shanghai Telecom's room can not tell Unicom's routers approached by, only in accordance with the Unicom router set a good route to go. Originally, even if the trip to Beijing does not have a big impact, after all, is the speed of light, but because most of Unicom's users to visit the telecommunications network is so around, Unicom's total exports become a bottleneck, the total export flow is too large, Unicom's users to visit the network speed will be slow. BGP line is no magic, but it can decide what route to go, do not take the road, the problem is naturally solved. It has the privilege of not only solving the problem of North-South interoperability, but also solving the interoperability problems of other networks, such as the Education Network. Because has the authority to decide the route, can optimize the route, which road block, I will change the path.
Iv. on-demand increase/decrease in bandwidth
Bandwidth is the primary cost, and when hosting a server, the bandwidth must be determined before signing a one-year contract with ISP. After a period of time, you find that the bandwidth is much more, want to reduce some is not allowed. To temporarily increase the bandwidth for a period of time is also not possible, to buy a year to buy (this is based on our contact ISP service provider). Therefore, generally will buy some more bandwidth, leaving some margin.
The use of cloud server can be flexible to increase or decrease the bandwidth, will not waste bandwidth, even if you buy less without worry, can be increased at any time. Although each cloud service provider will have certain restrictions, such as in Aliyun at least 1 months to buy the bandwidth, but more flexible than their own hosting server, the same bandwidth conditions, will save a lot of costs, especially the bandwidth requirements in a year changes in the larger site.
v. More attractive cost-payment methods
In the IDC room, the hosting server is usually a one-year contract, one-quarter fee paid.
The use of cloud services, can pay a shorter time, such as Aliyun can pay only one months of cost, saving liquidity.
---
On the whole, the almost cost, but has more memory, more CPU, more hard disk space, better bandwidth lines, more importantly, you can expand the computing resources on demand at any time.
That's why we choose cloud computing. |
