Explore Windows Azure failover mode and high-availability modes!

??

At present, many domestic users of the availability of cloud services there is misunderstanding, what kind of misunderstanding? For example, a cloud service provider, there is a room in South China, there is a computer room in east China.

The customer mentions a requirement, what do you mean by the concept of 99% usability? is not my machine in the Southern engine room out of the problem, my machine automatically transferred to East China computer room?

From the current communication and communication with customers, it seems that most users have this idea, that cloud services should be from cross-regional and cross-site direction of high availability, but this is a difficult to achieve the goal.

In the financial sector, where the concept of the three centres is often present, we can often see a description of the following definitions in the three centres of the two places:

Primary Data Center

All business and data hosting centers, in order to guarantee more reliable data access and high availability, the data center is generally located within a kilometer of our office location. All data, database, Access layer applications are hosted in the data center

Secondary data Center

The service of the secondary data center is equivalent to all copies of the primary hosted data center, in order to ensure the overall security of the data, our data Center is approximately about a kilometer from our primary data center, preventing small-scale disasters like fire and flooding in the primary data center.

The secondary data center makes data switching after a similar small-scale disaster in the primary data center, most of which use the secondary datacenter as an alternate site, and many enterprises use the auxiliary data center in the same city as their standby site.

When an alternate site is enabled, most users of the enterprise's Business data center are able to work with the office facilities of the secondary datacenter.

Secondary data centers in the third place

Usually the third data center is the same city data center because of the large-scale disaster caused by the inability to resume business in the near future, it is necessary to enable the third-place data center to start the business and ensure the continuous operation of the business. Usually the offsite data center recovery operation needs to do some network, the business and the data switch takes the time, therefore switches the third place the data center to need about 4 hours of time, after the switch, the key business personnel can go to the remote office through the airplane way, guarantees the vital business data to be able to continue to operate.

So most of the cases we mentioned in SLA 99% above in most cases are all local high availability conditions.

And based on the three-center model we are relatively invested is also very large, because it must be considered to reach the corresponding level of SLA needs we have to invest the corresponding hardware and software costs, for the two places, to achieve the equivalent level of application, we have to ensure that at least our hardware investment is 2 times, and when the third center is enabled, our cost is also the corresponding hardware cost is at least 3 times or more than dozens of times times.

The high availability that most cloud offers today does not mean that higher-availability SLA levels can be transferred across geographies, and in most cases high availability for on-premises applications, and remote high availability means that they are not implemented with a single-instance architecture.

At present, Microsoft Cloud has two different data centers in East China and North China, currently provides high availability of 99.95% , of course, we must also design a highly available architecture to ensure the normal operation of our data. There is a problem, when the data and application problems, how can we ensure that the data is normal? That depends on one of our key Azure features :

Trafficmanager

Traffic Manager Chinese translation for flow control, this lets many people think is the control traffic size, actually traffic Manger A very important function is to control our traffic trend.

Currently traffic supports three traffic trend modes:

2. failover

   2 cloud Services or a number of cloud services to failover, the system built-in automatic detection can detect the corresponding site and directory whether the situation, if there is a problem, he through the mechanism of automatic detection of traffic guidance to the normal site access.

3. Based on performance

   To load balance endpoints that are distributed across different datacenters around the world, you can direct incoming traffic to the closest endpoint because there is a minimum delay between the client making the request and the endpoint. Typically, the closest endpoint corresponds to an endpoint with the shortest geographic distance. You can use the performance load balancing method to distribute based on location and latency, but you cannot consider real-time changes in your network configuration or load.

   "Internet delay table " . The Traffic manager infrastructure runs tests to determine the round-trip time between the global differences and the data center that hosts the endpoint.

5. Dnsrr

   This way, users are given access based on the weight of the app being accessed, or the user's random access is implemented under the same weights, which is the load that is accessed based on the DNS Round Robin approach.

Our next goal is to implement failover of application access. We then implement load balancing for applications and access based on the simple complexity of the application.

Our next example will be a simple static page, based on SQL Azure and SQL virtual machines. There is also a multi-site synchronization of MySQL and MySQL virtual machines to complete the data's multisite access and replication technology.

Our next article will be configured as a simple way to operate and configure the failover model to meet different needs.

Copyright NOTICE: This article for Bo Master original article, without Bo Master permission not reproduced.

Explore Windows Azure failover mode and high-availability modes!