For decades, the IT environment has been dominated by the undisputed architecture of disaster recovery as a business continuity. But IT also awakens the ineffectiveness of this approach. All computing resources of the enterprise will remain idle until disaster occurs, and if this happens, all operations of the enterprise will be interrupted and will soon resume to a certain time. This method is no longer applicable to all times of all - weather access to all services. In this era, people have zero tolerance for downtime. Many online business companies say that the interruption of network services will cause huge losses to its revenue and cause significant losses to its reputation. Today, enterprises need to build their own infrastructure to achieve continuous availability, instead of taking time and energy to build a disaster recovery structure.
With the application and development of virtualization, big data, cloud computing and modern database management system, double live data center architecture is no longer an option, but also a necessary condition to ensure continuity of data center business. But they also bring a series of new challenges, because they achieve high cost, complex construction, extra equipment and low latency. These architectures are particularly difficult to build on the data layer. Fortunately, database load balancing software makes the architecture of double live database operations more simple and lower cost. The traffic management function of the load equalizer enables the enterprise to reduce costs and reduce downtime by using the following four features:
(1) automatic support for read / write separation. The first step of the double live operation is to use multiple servers to support database reading. The challenge is to rewrite the application to send the read to the auxiliary node. The database load balancing software solves this problem by automatically pointing the written transaction to the main server and reading it to the auxiliary server. Automatic read / write separation can provide instant application performance and reliability gain without the need to modify the application.
(2) automatic fault switching. The load balancer uses a modern database built-in fault transfer mechanism to enable a transparent fault handoff of the application. Database performs replication and failover. Database load balancing software saves inbound requests during failover, so applications will see delays but no errors.
(3) enable the geographical location of the load balance. The database load balancing software keeps measuring the time of the first byte and directs traffic to the fastest server. It can also balance the reading load of all available auxiliary parts to provide fast application performance.
(4) replication awareness monitoring. The design of the database load balancing software is designed to monitor the replication delay consistently and determine the nodes that are lagging behind the predetermined threshold. Then avoid sending the request to the expired server, so it does not provide outdated data.
Compared to disaster recovery, the double live operation is more expensive, but only about 20%, while providing 35% capacity and uninterrupted operation. This increase cost more than the increased capacity cost, extended the normal running time, improved performance and the best asset utilization. When a database load equalizer is deployed in a continuous available architecture, they can also be maintained at zero downtime.
This new breakthrough technology is enabling enterprises to avoid interruptions and ensure continuous operation by changing from disaster recovery to sustainable availability.
Database load balancing software is the most reliable and economical way to build a double live data center. No code changes are needed, and even in the most harsh conditions, the load equalizer can provide significant performance improvements. With the help of inter memory query cache and application transparency and extensibility, it can bring maximum value to online business, no matter in internal operation or cloud processing.
沒有留言:
張貼留言