Details
Both of our clustered plans share the same feature
sets. Each node (machine) in each cluster participates in serving and
processing parts or the complete web site.
Using a proprietory processing engine, the next server is selected by Round
Robin, Fastest Response, or a Static Weight algorithms. In
addition, each server is assigned has a sticky attribute (SA).
SA enables the detection of sessions and holds nodes resposible for
the entire client sessions. For programmed sites with server-side session
variables, scaling to our clustered solution requires no code rewrites and
zero impact to live sites.
In addition to
proactive failover, our cluster configuration reads site data from
header strings <HEAD>, and determines site availablity for
every node. If any probe fails, an notification email is sent
to the administrator and the node is dropped out of the cluster
until it resumes to serve the specified <HEAD> string.