More content, add number of public attention:
scene one:
After a smart home manufacturers, users shouted "Little X classmates, help me to sweep the floor," the server is down, the cleaning robot can not be started immediately, so users may call several times after another in a row, helpless and habits pressed cleaning robot the start button.
Scene Two:
A high-rise building has 2000 rooms, 10 rooms smoke alarm in a row, in theory, there has been a fire and gradually spread, as it happens, and the server is down, then ...
Two scenes are server goes down, but the consequences are different, based on actual business situation, we must consider the high availability software architecture.
Some would say it on the cloud, the cloud server more stable than build their own. Typically this is the case, but do not overlook this important concept of SLA, SLA's cloud offerings are there, what is it SLA? SLA stands for ServiceLevel Agreement, translated into service level agreements, he shows a public cloud service level and quality. For example, we say that 99.95% monthly SLA, monthly service time means a failure only 0.05% of the total time, if this month is 30 days, equal to about 21.6 minutes.
For a scene, perhaps 21.6 minutes is tolerable, but the second scenario is definitely not allowed.
High availability for a variety of programs, usually referred to the following concepts (not the concept for Azure's):
City disaster recovery
Disaster recovery is in the city or near the city area (≤ 200K M) establishing two data centers: one for data centers, responsible for the daily production runs; the other is a disaster recovery center, is responsible for running the applications after a disaster. City data center disaster recovery and disaster recovery center distance of more recent, better communication line quality, relatively easy to implement synchronous data replication, to ensure high data integrity and zero data loss. City disaster recovery are generally used for fire prevention, building damage, disaster power failure, computer systems and vandalism caused.
Disaster Recovery
The distance between the primary site disaster recovery center far (> 200KM), off-site disaster recovery can not only prevent fire, building damage and other potential risks that may be encountered, but also to prevent war, earthquake, flood and other risks. As the city off-site disaster backup and disaster recovery own strengths, to achieve the best effect of disaster prevention, data centers should consider the establishment of a city and off-site disaster recovery center each way around.
There are a lot of high-availability solutions that we can choose, in order to maximize our business system as stable as possible to provide services on Azure, first thing to do is to understand the concept of high availability on the Azure related to flexible application.
What is Azure area?
Azure run in multiple data centers around the world. These data centers are grouped into geographic regions, allowing users the flexibility to choose the location to build the application. Can ( "2 in eastern China," such as "in eastern China", "China northern" or) create Azure resources in the specified geographical area. In order to provide redundancy and availability, each region has a plurality of data centers. In this way, you can flexibly design application, create closest to the user's VM, meet any legal, tax or compliance requirements. As of November 2019, Azure a total of 54 regions of the world, there are four areas in mainland China.
Azure Azure geographical area in the form of regional organizations. Azure ensure geographical data resides, it requires sovereignty, compliance and recovery capabilities to be complied with within a geographical boundary.
Area of
Each Azure regions paired with another within the same geographic region. This method is suitable for replication across geographies resources (such as VM storage), reducing natural disasters, social unrest, power outages or physical network outage affects both the possibility of the two regions. Other advantages of the region comprising: when a wide range of Azure interrupted region, each region of the region has a higher priority, which can shorten the reduction time of the application. One by one to be paired region Azure updates planned, to minimize downtime and reduce the risk of application downtime. For tax and jurisdictional law enforcement, the data still resides in the same form paired geographical area.
Azure Overview of High Availability
Azure can create high availability through a variety of different ways. There are two options are "set availability" and "availability zones." By using a set of availability, your VM will be protected from the effects of the failure so that might occur within the data center. This includes hardware failures and software failures Azure. By using availability zones, your VM will be placed in a physically separate, no shared infrastructure resources, it will be protected from the impact of the entire data center failures. When you want to deploy reliable solutions based VM in Azure, use or availability availability set area.
Availability Sets Overview
Availability is a logical grouping feature set, it can be used in the Azure VM resources to ensure that the data center is deployed on Azure, these resources isolated from each other. Azure deployed in the VM ensure availability across multiple physical servers, computer racks, the storage unit and the network switch operation. Azure if the hardware or software failure, VM will be affected only part of the overall application will keep running, available for customers to use. If you want to build cloud solutions and reliable availability is a key feature set. Suppose a typical VM-based solution includes four front-end Web server, and two back-end VM that hosts the database. In the Azure, if you want before you deploy VM availability to define two sets: a set of availability for the "Web" level, another set for the availability of "database" level. When you create a new VM, you can specify the availability set in azvm create command as an argument, Azure automatically ensures that remain independent across multiple physical hardware resources in the VM availability of centralized created. If you are running a Web server or database server's physical hardware problem, you can be sure other examples of Web servers and database VM will keep running because they are located on different hardware.
Availability Zones Overview (as of November 2019, in China there is no availability zone)
Availability zone is a high availability products / services in the data center fails to protect applications and data. Azure availability zone region is unique physical location. Each region consists of one or more data centers, data center configuration independent power supply, cooling, and network. To make sure you can restore all areas enabled must have at least three separate areas. Physical isolation region in the region of the availability of applications and data can be protected in the event of failure of a data center. Redundant service area across availability zones copy applications and data, to protect against single points of failure. With the availability of the region, Azure provides the industry's best 99.99% VM uptime SLA. And the availability of similar sets, suppose a includes four front-end Web servers, database hosting and back-end VM is based on typical two solutions VM. Similar sets and availability, you will want to deploy two separate availability of the VM region: a region available for "Web" layer, another area for the availability of "Database" layer. When you create a new VM and availability zones designated as azvm create command parameters, Azure VM will automatically ensure that you create completely isolated in different availability zones. If you are running a Web server or an entire data center database server VM is a problem, you'll know other examples of Web servers and database VM is still running, because they run completely independent data center.
Azure availability zone region is a combination of fault-tolerant update domain and domain. For example, if three or more to create three regions Azure VM region, the VM effective fault-tolerant distributed in three fields and three fields updated. This distribution is on the Azure platform will recognize the updated domain to ensure that different regions of the VM is not updated at the same time. By calculation, storage, network resources and co-location data and copy it to other regions in a region, in the built-in high-availability application architecture. Support availability of Azure service area is divided into two categories:
Regional services - fixed resources to a specific area (for example, a virtual machine, hosted disk, standard IP address) or
Redundant service area - platform automatically replicated across the region (such as redundant storage area and SQL database).
To achieve a comprehensive business continuity on Azure, Azure using availability zones and regions to build on the application architecture can be combined. You can use Azure availability zone in the region of copying applications and data in a synchronized manner to achieve high availability and asynchronously replicated across the region Azure for disaster recovery protection.
Fault-tolerant domain
Fault-tolerant hardware logic domain is a basic set of common shared power and network switches, similar to rack within the local data center. When you create a VM in the availability of collection, Azure VM platform will automatically fault-tolerant distributed to these domains. This approach may limit the potential for physical hardware failure, network outage or power failure affected.
Update fields
Update the underlying hardware domain is a logical group can simultaneously maintain or restart. When you create a VM in the availability of collection, Azure platform will automatically distribute these updates VM domain. When the Azure platform for regular maintenance, this approach ensures that at least one instance of the application running at all. During the planned maintenance, update the field to restart the sequence may not be sequential, but only once a reboot to update the field.
Availability set consists of hardware failures can be prevented, and two additional update packets allow secure applications - fault-tolerant domain (FD), and update domain (UD).
Understand these concepts, let's walk through each load balancing on Azure, application gateways, traffic management and other products, then in actual combat operations, see if you can unplug the final production environment network cable.