eCluster™ 4.x

• Load Balancing and Fail-Over
• Geographical Proximity Load Balancing
• Content Aware Clustering
• Load Balance and Fail-Over Cluster As Service
• Load Balancer and Software Load Balancer
• Success Stories
• Feature Details

July 1, 2017, AiForce announced the shipment of eCluster™ 4 designed for Enterprise and ISP/Cloud provider.  The eCluster™ 4, with our ground-breaking CFS™, a Clustered File System, is by far the most advanced clustering solutions for UNIX and Windows in the market.

The eCluster™ Internet Clustering Technology is characterized by:

• Unprecedented High Availability
• Unprecedented Scalability
• AiForce Dynamic Server Feedback Protocol - Cluster Agent
  

The product is designed to cluster Internet/Intranet Servers into Virtual Cluster Groups (supported up to 1 Million Virtual Groups - Virtual Domains).  Each Virtual Cluster Group supports up to 1024 Cluster/Cloud Nodes, where server load balancing, fail-over, file system clustering, and SQL clustering are made possible for both servers and workstations.  With eCluster™, ISP may load balance and failover among several geographical locations internationally.  By load balance and failover each geographical location, or mirroring between several geographical locations, ISP will be able to eliminate downtime completely, and even being able to provide clustering services remotely for Web Hosting customers.

Load Balancing and Fail-Over
The eCluster™ load balancing ensures that network traffic will always go to the server (Cluster Node) with lowest load average, memory usage, user logins, and network connections, etc.  A heavily used/congested server will slow down CPU processing speed.  When these processing loads (jobs) are distributed evenly among servers (Cluster Node), CPU task can be finished sooner and faster even if the load is high.

Fail-over ensures non-stoppable CPU processing, even under a catastrophic disaster such as building power failure, fire, earthquake, where the server is physically damaged.  Power generator and UPS is not needed any longer.  With 2 or more eCluster Device/System running from 2 or more geographical location, even internationally, if one data center is destroyed (long hours of data restoration is needed), other data centers will be able to take over instantly.

In a single geographical location setting, the eCluster™ device/system can ensure non-stoppable CPU processing among all of it's Server Node.  If one node goes down for any reason (such as HTTP crash/malfunction, OS crash, system upgrade), the other node will take over automatically.  One may have 2 or more eCluster™ master controller device/system running to backup and load balance each other to ensure if one eCluster™ controller goes down, the second one takes over.

Geographical Proximity Load Balancing
The eCluster™ load balancing detects how close the client is to each of the Clustered Servers in a Cloud geographically by comparing client location and the server's location, then route the client to the closest Server Node with least load. 

Example One: 2 servers, Server A is located in New York, city A. The Server B is located in New York, city B.  The Internet User is located in New York city B. Therefore, the Internet User will be routed to the New York city B's Server B.  This greatly improves web user experience since the closest server is always the fastest to user.

Example Two: 2 servers, Server A is located in New York. Server B is located in California.  The Internet User is located in New York. Therefore, the Internet User will be routed to the New York Server A, instead of Server B in California.  This greatly improves web user experience since the closest server is always the fastest to user.

Example Three: 4 servers, 2 Servers A are located in France.  2 Servers B is located in USA.  The Internet User is located in USA.  Therefore, the user will be routed to access the 2 USA servers with load balanced which is closest to him or her, instead of the France servers.

Content Aware Clustering
URL Fail Over, Cold Fusion Fail Over, SQL Fail-over - sometimes the front end Web Servers are running OK, but it's backend or middleware may be failed.  With Content Aware Clustering, eCluster™ is able to detect if URL, Cold Fusion, or backend SQL database are capable to support their front end Web Servers, and switch traffic only to the best functioning Cluster Node.

Load Balance and Fail-Over Cluster As Service
The eCluster™ Clustering and Load Balancing Technology enables ISP/Cloud provider to provide server clustering service for their web hosting customers as a value added service.  Network traffic can be distributed in distant computer centers across the world with failover assurance.

1. Clustering Each Hosted Domain with Smart Load Balance and Fail-over Capability - The eCluster™ supports the clustering of over a million of ISP's hosted Web sites (domains).  Each clustered Web site (domain) may contain up to 1024 cluster nodes or more.
2. Smart Load Balance and Fail Over - eCluster™ load balance and fail over according to system load average, CPU usage, number of network connections, memory usage, number of processes, etc.  It's able to load balance across computer centers located in different countries and continents. For example, if all 4 nodes in the computing center at LA have load larger than the 8 nodes in Japan, the load will be migrated to Japan's computing center.
3. 100% Uptime - With failover between several computing centers, a 100% uptime can be achieved.  For example, if one of the 4 nodes in LA is down due to one of the followings:

• IIS/Apache crashed, or
• OS crashed, or
• Software/Hardware upgrade - swap in a bigger hard drive.

One of the lowest loaded nodes from the rest 11 nodes (3 nodes in LA and 8 nodes in Japan) will take over the functioning of the offline node.  If any one of the computing centers is totally down, such as local power plant failure or earthquake, the other computing center will be able to take over.  Administrators may feel free to turn off one or several node in a cluster group (not all nodes) for hardware maintenance, not worry about OS and/or IIS/Apache crashes.

A Canadian ISP has successfully provided load balancing and fail over web hosting service for several leading casinos.  They provided a 20 cluster group with 10 cluster node for their hosting customers.  Traffic is distributed in several computer centers across North America with failover assurance.

Click the "Product Tree" on the left to learn more about our eCluster™ product.

Load Balancer and Software Load Balancer
Both load balancer and software cloud clustering are available for various Operating Systems. 

Success Stories
For success stories, click here.