Internet IronWare's global server load balancing transparently expands load balancing
capabilities to globally dispersed data centers and server farms. This gives network
managers the flexibility to easily direct traffic to the most appropriate server or server
farm based on criteria such as server health, load, proximity or average round trip delay
without regards to the physical location of the server. This capability provides high
resiliency by allowing network managers to locate fail-over servers in different
facilities for disaster recovery.
ISPs and enterprises can use global server load balancing to build highly scalable and
fault tolerant global services and applications. Global server load balancing enhances end
users' perceived performance and reduces bandwidth costs by directing users to the nearest
data center.
Enhanced Service Redundancy, Availability and Performance
Internet IronWare and the ServerIron platform ensure service availability by offering
switch, server, link and session level redundancy.
In the event of a server or server application outage, ServerIron provides sub-second
detection and fail-over to the next server in a logical group that supports a like
service. This ensures that traffic continues to flow and applications are available to
end-users.
To provide 100 percent uptime of critical services, ServerIron includes a hot stand-by
redundant switch capability that protects against session loss. With this feature,
ServerIron allows primary and secondary switches to support identical configuration
parameters. The secondary switch continuously monitors the traffic being serviced by the
primary switch. If the primary switch fails, the secondary switch takes over without
losing sessions or connectivity. ServerIron also offers link level redundancy that ensures
server connectivity by automatically switching sessions from a failed link to a redundant
link.
Internet IronWare's service health check capability monitors servers' ability to
respond to user requests by determining whether an HTTP, DNS, Radius or SMTP service has
failed, in addition to monitoring server connectivity. This ensures fast detection of
service problems and eliminates the risk of service outage.
Symmetric server load balancing provides up to 2,000,000 connections and full service
redundancy to mission critical services. With symmetric load balancing, multiple parallel
ServerIrons share active loads and act as a backup to each other in the event of a
failure.
Direct server return supports up to 1,000,000 concurrent active user connections and
enables requested content to be served through the best return path. This significantly
improves end-users' performance and allows system performance to scale based on the
requirements of applications such as web pages, graphics and streaming media.
High Performance Content Publishing and Web Hosting
Internet IronWare provides support for an unlimited number of VIPs. With this feature,
network managers can configure thousands of VIPs with a simple set of commands. Web
hosting companies can use this feature to easily support hundreds of home pages that are
identified as discrete domain names and IP addresses and then load balanced over multiple
servers.
With ServerIron, network managers can utilize URL switching to reduce content
replication costs and management overhead. ServerIron inspects each client's URL request,
for example www.foundrynetworks.com, and sends requests to a specific server that hosts
the requested content. Thousands of URLs are supported and content can be load balanced to
local or remote servers.
Internet IronWare's many-to-one load-balancing feature enables network managers to
easily track service usage by VIP. Typically, there is only one VIP to server association
per TCP/UDP port. With this capability, a single server and port number can service
multiple VIPs and compile extensive usage information for each VIP. Information such as
site visits, usage statistics and other service-related data is crucial to ISPs in
sustaining and growing an Internet web site.
Improves Internet Response Time and Reduces WAN costs with Transparent Cache
Switching
ServerIron's transparent cache switching feature improves Internet response time and
reduces WAN operating costs by redirecting web traffic destined for remote Internet hosts
to a group of local cache servers. Foundry's transparent cache switching can be used with
any cache server that supports transparent re-direction, including those from leading
vendors such as CacheFlow, Cobalt, Inktomi, Network Appliance and Novell.
Foundry's transparent cache switching offers network managers a resilient web caching
solution that significantly simplifies administration. Unlike proxy server solutions that
require manual configuration of each clients' browser, ServerIron transparently intercepts
and switches HTTP client requests to an available cache server. No client browser
reconfiguration is required. Network administrators can configure ServerIron to switch
traffic based on source and destination IP address.
Transparent cache switching also provides network managers with a level of redundancy
and resiliency that is unavailable with proxy servers. With ServerIron, managers can
deploy redundant web cache server configurations, known as web cache server groups, that
provide automatic recovery in the event of a failed web cache server. This feature ensures
that critical web traffic continues to flow across the network at all times. For
additional resiliency, redundant ServerIron switches can be deployed to eliminate any
single point of failure to a web cache server group.
Cache Route Optimization (CRO) adds intelligence and further reduces response time by
forwarding cache return packets to the most appropriate router. CRO detects packets that
are routed inefficiently, which can cause significant performance degradation at the
Internet router and poor service response time to clients. CRO leverages the ServerIron's
ability to process packets at wire speed. By examining each packet's Layer 3 and Layer 4
headers and being application aware, the ServerIron correctly switches the packets
generated by the cache servers to clients without intervention by the default Internet
router. With CRO, ServerIron reduces traffic load at the Internet router and significantly
improves user response time.
In some network environments, cache servers receive traffic from a router based on a
configured policy; however, the router has no mechanism to detect a failed cache server.
With Internet IronWare's Cache Failover Protection (CFO), ServerIron ensures service
availability in the event of a cache server failure. A cache health check detects service
failures and redirects the client traffic to the Internet as soon as the cache failure
occurs.
Policy-based cache switching gives users the flexibility to direct client requests to
the appropriate cache server based on service policy and destination. With this feature,
users that request specific web sites are directed to special content caches that include
pre-loaded content. Policy-based cache switching allows publishers to distribute content
in anticipation of peak content demand driven by current events.
Direct cache server return amplifies cache return bandwidth to improve overall service
performance and customer satisfaction. Direct cache return enables cache servers to more
efficiently serve content by taking a higher bandwidth return path than that taken by the
original request.
Service Security for Internet Integrity
For increased network security, ServerIron can be used to restrict access to specific
applications from a given address or sub-net. Filters can be easily set to deny access to
servers on a port by port basis. For example, a network administrator can deny FTP traffic
to a particular address. Conversely, filters can be set to allow access by a subset of
users or sub-nets.
ServerIron also protects against malicious TCP SYN attacks by allowing administrators
to configure a threshold for the rate of incoming TCP SYN packets. If the threshold is
exceeded, ServerIron automatically stops binding new sessions for a period of time.
Layer 4 QoS/CoS Prioritizes Application Traffic
ServerIron can be used to effectively prioritize TCP and UDP traffic. This capability
ensures that critical traffic destined for servers that support multiple applications can
be given priority over other traffic destined for the same servers.
With ServerIron, packets are prioritized based on a combination of destination address
and destination port number. This allows network managers to provide critical applications
with a higher class of service.
Accounting Statistics Improve Management of Application Traffic
ServerIron's accounting statistics allow network managers to easily collect and display
detailed information about network traffic destined for server farms. The amount of
traffic between servers and clients, as well as which application is dominating network
traffic, can be tracked. The number of active and open sessions on servers can also be
monitored. These statistics can be used to track traffic loads on servers that support
multiple applications.
Configuring and Managing ServerIron
ServerIron can be configured and managed using Foundry's IronView network management
tools. Simple Network Management Protocol (SNMP) device management and configuration
applications are available on major platforms, including HP OpenView for Sun Solaris and
Windows NT and standalone Windows NT. Web/HTTP is available, as is a Command Line
Interface (CLI) for local and remote management and configuration. Remote Monitoring
(RMON) and a mirror port provide network monitoring and troubleshooting.
