Load balancing improves performance of the scalable service, both in response time and in throughput. There are two classes of scalable data services.
A pure service is capable of having any of its instances respond to client requests. A sticky service is capable of having a client send requests to the same instance. Those requests are not redirected to other instances.
A pure service uses a weighted load-balancing policy. Under this load-balancing policy, client requests are by default uniformly distributed over the server instances in the cluster. The load is distributed among various nodes according to specified weight values. For example, in a three-node cluster, suppose that each node has the weight of 1. Each node services one third of the requests from any client on behalf of that service. The cluster administrator can change weights at any time with an administrative command or with Sun Cluster Manager.
The weighted load-balancing policy is set by using the LB_WEIGHTED value for the Load_balancing_weights property. If a weight for a node is not explicitly set, the weight for that node is set to 1 by default.
The weighted policy redirects a certain percentage of the traffic from clients to a particular node. Given X=weight and A=the total weights of all active nodes, an active node can expect approximately X/A of the total new connections to be directed to the active node. However, the total number of connections must be large enough. This policy does not address individual requests.
Note that the weighted policy is not round robin. A round-robin policy would always cause each request from a client to go to a different node. For example, the first request would go to node 1, the second request would go to node 2, and so on.
A sticky service has two flavors, ordinary sticky and wildcard sticky.
Sticky services enable concurrent application-level sessions over multiple TCP connections to share in-state memory (application session state).
Ordinary sticky services enable a client to share the state between multiple concurrent TCP connections. The client is said to be “sticky” toward that server instance that is listening on a single port.
The client is guaranteed that all requests go to the same server instance, provided that the following conditions are met:
The instance remains up and accessible.
The load-balancing policy is not changed while the service is online.
For example, a web browser on the client connects to a shared IP address on port 80 using three different TCP connections. However, the connections exchange cached session information between them at the service.
A generalization of a sticky policy extends to multiple scalable services that exchange session information in the background and at the same instance. When these services exchange session information in the background and at the same instance, the client is said to be “sticky” toward multiple server instances on the same node that is listening on different ports.
For example, a customer on an e-commerce web site fills a shopping cart with items by using HTTP on port 80. The customer then switches to SSL on port 443 to send secure data to pay by credit card for the items in the cart.
In the ordinary sticky policy, the set of ports is known at the time the application resources are configured. This policy is set by using the LB_STICKY value for the Load_balancing_policy resource property.
Wildcard sticky services use dynamically assigned port numbers, but still expect client requests to go to the same node. The client is “sticky wildcard” over pots that have the same IP address.
A good example of this policy is passive mode FTP. For example, a client connects to an FTP server on port 21. The server then instructs the client to connect back to a listener port server in the dynamic port range. All requests for this IP address are forwarded to the same node that the server informed the client through the control information.
The sticky-wildcard policy is a superset of the ordinary sticky policy. For a scalable service that is identified by the IP address, ports are assigned by the server (and are not known in advance). The ports might change. This policy is set by using the LB_STICKY_WILD value for the Load_balancing_policy resource property.
For each one of these sticky policies, the weighted load-balancing policy is in effect by default. Therefore, a client's initial request is directed to the instance that the load balancer dictates. After the client establishes an affinity for the node where the instance is running, future requests are conditionally directed to that instance. The node must be accessible and the load-balancing policy must not have changed.