Intergraph WebScale User Manual page 54

WebScale™ User's Guide
How WebScale Works
WebScale enables all cluster hosts on a single subnet to simultaneously listen to
incoming network traffic for the cluster's primary IP address (and for additional IP
addresses on multi-homed hosts). On each cluster host, the WebScale driver acts as a
filter between the cluster NIC driver and the TCP/IP protocol stack to allow a portion of
the incoming network traffic to be received by the local host. With this capability,
WebScale can efficiently deliver both high availability and scaled performance.
WebScale delivers high availability by redirecting incoming network traffic to working
hosts as necessary after a host fails. Outstanding connections to a failed host are lost,
but the Internet service remains available. In most cases (for example, with web
servers), client software automatically retries the failed connections, and the clients
experience only a few seconds delay in receiving a response.
WebScale delivers scaled performance by partitioning the incoming network traffic
among multiple hosts in the cluster so that these hosts can simultaneously respond to
different client requests, even multiple requests from the same client. For example, a
web browser would obtain different images for a single web page from different hosts
within a WebScale cluster. This speeds up processing and shortens the response time to
clients.
WebScale controls the distribution and partitioning of TCP and UDP traffic from the
Internet clients to selected hosts within a cluster as follows. After WebScale has been
configured, incoming client requests to the cluster IP address(es) are received by all
hosts within the cluster. WebScale filters incoming datagrams to specified TCP and
UDP ports before these datagrams reach the TCP/IP protocol software. WebScale
manages only the TCP and UDP protocols within TCP/IP, and its actions are controlled
on a per-port basis.
WebScale does not control any incoming IP traffic other TCP and UDP traffic for
specified ports. It does not filter ICMP, IGMP, ARP (except as described below) or
other IP protocols. All such traffic is passed unchanged to the TCP/IP protocol software
on all of the hosts within the cluster. Because of TCP/IP's robustness and its ability to
deal with replicated datagrams, other protocols behave correctly in the clustered
environment. However, you should expect to see duplicate responses from certain
point-to-point TCP/IP applications (such as ping) when the cluster IP address is used;
these applications can use the dedicated IP address for each host to avoid this behavior.
To coordinate their actions, WebScale hosts periodically exchange multicast or
broadcast messages within the cluster. This allows them to monitor the status of the
cluster. When the state of the cluster changes (hosts fail, leave or join the cluster, etc.),
WebScale invokes a process known as convergence, in which the hosts exchange
messages to determine a new, consistent state of the cluster and to elect the host with
the highest host priority, known as the default host. When all cluster hosts have reached
consensus on the correct new state of the cluster, they announce the completion of
convergence in the Windows NT event log.
50