Load Balancing vs Clustering

Discussing the differences between the cluster and load balancing, if one or more ago, you should download the definitions on the right. Clustering solution is used as a synonym for third-party software-based devices and is understood to mean a lot to offer load balancing and load balancing capabilities of the software often several.

In practice, clustering, in general, IBM WebSphere, BEA WebLogic and, Oracle AS (10 grams) are used in applications such as servers. Also used in such an environment, load balancing is Apurikeshonderibarikontorora (ADC) in, BIG in features such as IP. (For simplification, ADC will compare the story of the unified approach.)

Scalability, speak toward the horizontal

Of course, some Hadouearodobaransa, our swimming pools and farms, and distributed applications, which are talking about application server clustering. In the software world, the term cluster will be applied to the same group.

Integration is usually the master controller, the process will be converted to an instance of the application server / distributed requests to multiple instances of the algorithm, round robin, to use industry standards such as weighted round-robin or least connections. And horizontal scalability for clustering is to add additional application server instances to increase capacity and performance and response time is virtually seamless, which is similar to load balancing. Is actually used in the forums, usually, ICMP ping to determine whether the method used to secure the clustering, in some cases, HTTP or TCP connectivity checks.

Health and transparency

For load balancing, which can be, that other, more complex number-crunching, and server CPU and memory utilization, faster response time and test parameters, such as clustering solution to be simple server than the application of rigorous algorithms to support health monitoring, and supports the same area. This is what you can see them, passive surveillance, and providing a means of checking the health of low-load application server instance.

Applications, the conference is the affinity to interact with the same server, server clustering using the user name is required. This information, if it is common for entry during the process, the data session page, such as basketball, you need to store the completion of the transaction (request) and used between.

For the same situation, the persistence to use the CAN. A, C, as well as applications and data from the network to get information about other applications that use the traditional variable clustering is usually not, very, available variables are limited.

Node agent over several clustering solutions are clustered by the controller, you must have your application server deployed on each instance.として多くの場所にあるが、展開と管理の程度に大きな負担にはならないかもしれないが、さらにはプロセスのリソースをCPUサーバとメ モリ消費量を実行している。 Of course, it also adds another point of potential failure of the data path. Since, C does not require a server-side components, they remain completely transparent.

Make the right choice

There are some questions: such as a database server to another when multiple servers, Web servers, as needed, such as messaging, a distributed system can play a particular role, software, server and cluster configuration to build the extra work Why?

How do you choose how? This is the first study that is why this type of solution is (perhaps) to vary the application server clustering capabilities that are specific to whether to purchase additional. Also, whether (or) is the broader question of the need to provide multiple application server support the brand. Clustering, of course, is its own application server, and any and all CAN application or Web service may be provided to the server.

Clustering Checklist

Features:

* Package enterprise server applications and typically

* Network does not require the highest level of expertise

* Following a lengthy deployment usually more expensive than ADC

Weaknesses:

* High availability solutions, clustering is not guaranteed

* Best Practices for Deploying a cluster controller on another machine

* The node agent is required to manage the application on the server instance

* Clustering "owner" of the same type of server (which is the only group you)

ADC Checklist

Features:

* Provide a heterogeneous environment, high availability and load balancing

* Application optimization, security, and acceleration values

* You need to be deployed without changing applications or server

Weaknesses:

Works of architecture and infrastructure *

* More expensive than many clustering solutions typically

* Must be set in May to expand the new technology / management

Recommendations

ADC test by reading the cluster of servers based on several articles based on tests and exams, performance, configuration, get a good view case studies. Look for case studies, as much as possible, mirror the situation and if your (or thinking) people believe what you are talking to people. Contrary to the government, and entered the automotive and health care, and not do something soon, something to please him. Please note this decision.