Multiplex communications – Communication techniques for information carried in plural... – Adaptive
Reexamination Certificate
1999-06-30
2003-12-02
Cangialosi, Salvatore (Department: 2732)
Multiplex communications
Communication techniques for information carried in plural...
Adaptive
C709S222000, C709S241000
Reexamination Certificate
active
06658018
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention generally relates to the field of data communication networks. More particularly, the present invention relates to a method and system of providing advanced teaming functionality, capable of accommodating heterogeneous adapters to improve the utility and performance of a fault-tolerant network adapter team.
2. Description of Related Art
Most computer communication networks designate at least one computer as its network server. Servers operate as centralized repositories for all the applications and work-files available on the network and provide an expedient way to interconnect and execute programs that are distributed across a network. Servers are also responsible for managing the access and flow of information on the network. As a result, enterprises are increasingly relying on such networks to support both routine and business-critical applications.
Such networks, however, are not without weaknesses. For example, as illustrated in the lower portion of
FIG. 1
(Prior Art), a particularly susceptible element in all networks is the network link
191
A to the server
100
A. The network link
191
A is physically coupled to an adapter
101
A (i.e., network interface card) installed within the server
100
A. As such, all data traveling to, and from, the network must pass through the adapter
101
A. A failing adapter
101
A will, therefore, compromise the network server
100
A connectivity and ultimately result in the loss of productivity.
In an effort to guard against such vulnerability, fault-tolerant adapter teams
105
, as shown in the upper portion of
FIG. 1
, have been developed which rely on redundancy for their restorative capabilities. These adapter teams
105
, typically comprise a set of M adapters
101
,
101
B,
101
C, . . .
101
M along with corresponding team software, and can be manually configured, by the network administrator, to designate one of the adapters as “primary”
101
and the remaining adapters as “secondary adapters”
101
B,
101
C, . . .
101
M. The primary adapter
101
is configured to support a primary connection
191
to the network and embodies the identity of the adapter team
105
by having its Media Access Control (MAC) address serve as the network adapter
101
address. As such, all communications destined for the server
100
shall be directed towards the primary adapter.
Meanwhile, the secondary adapters
101
B,
101
C, . . .
101
M serve as back-up connections. The adapter team
105
software contains one or more intermediate drivers
185
that monitor all network connections
191
,
191
B,
191
C . . .
191
M and, in the event of primary adapter
101
failure, the software provides “fail-over” capability by defaulting to a pre-selected secondary adapter (i.e.,
101
B) as the alternate primary. The pre-selected secondary adapter is the secondary adapter in the adapter team
105
determined to be most capable. Traffic is subsequently re-routed to the alternate primary adapter, while still utilizing the original primary (or team)
101
address. Thus, the pre-selected secondary adapter
101
B automatically assumes ownership of all data going to the server
100
, without the need for human intervention. In some instances, the team's software also provides for automatic “restore-back” capabilities, which allows a replacement adapter, positioned in a team slot previously occupied by a failing primary adapter
101
, to be re-instituted as the primary if the software senses that the adapter is fully operational. In this manner, the redundant set of adapters
101
,
101
B,
101
C . . .
101
M provides network link recovery that is generally transparent to applications and users.
It is also possible to capitalize on the existence of multiple adapters in a team to increase the effective system bandwidth. For example, Adaptive Load Balancing techniques make use of the team's M adapters
101
,
101
B,
101
C, . . .
101
M to increase performance by balancing the transmission of data from the server
100
to the network. In such a configuration, the team's address, which as stated above, is the primary adapter's address, is advertised to the network. Accordingly, all server-bound traffic is received by the primary adapter
101
. The remaining M-
1
team adapters (referred to above as secondary adapters)
101
B,
101
C, . . .
101
M do not receive any traffic originated by the client. However, the M-
1
team adapters
101
B,
101
C,. . .
101
M, as well as the primary adapter
101
, do transmit data from the server
100
to the network. The transmission of data is achieved by ratably allocating the data across the primary adapter
101
and remaining M-
1
team adapters
101
B,
101
C, . . .
101
M. This effectively aggregates the outgoing server bandwidth, thereby increasing network throughput and mitigating bottlenecking episodes.
Although conventional fault-tolerant adapter teams
105
are capable of providing fast and reliable recovery from adapter failures, a high premium is often paid in both hardware cost and network efficiency. One reason for such a high premium in conventional teams
105
arises from the need to have adapters of the same type or speed to ensure a seamless fail-over transition. This significantly increases the cost of system maintenance by forcing the purchase of similar adapters and rendering dissimilar adapters unusable. This also limits the choice of adapters that can be utilized in fault-tolerant teams
105
or Adaptive Load Balancing teams
105
. Such limitations can be particularly costly when changes in the server
100
configuration precipitates adapter upgrades, rendering all the inventoried adapters obsolete.
Moreover, conventional symmetric fault-tolerant teams
105
require the manual designation of the primary adapter
101
. These conventional teams
105
are, therefore, incapable of either automatically initializing the primary adapter
101
or doing so in an intelligent and efficient manner.
Furthermore, Adaptive Load Balancing teams
105
capable of accommodating dissimilar adapters, cannot select, “on the fly”, the adapter with the next-best performance profile that optimizes network throughput. Because the incoming data is solely routed to the primary adapter
101
and the outgoing data is proportionately balanced across all M team adapters
101
,
101
B,
101
C, . . .
101
M, a non-optimal secondary adapter
101
B selection can compromise the efficiency of a network.
In addition, some advanced servers
100
possess hot-plug capabilities, allowing network administrators to swap components without powering down the server
100
. With respect to hot-plug-compatible fault-tolerant teams
105
, hot-plugging is achieved by electrically isolating a port in the team while maintaining power to the rest of the team ports. Conventional hot-plugging adapter teams
105
, however, only permit the failed adapter to be replaced by an identical adapter. This further limits the choice of adapters available for fault-tolerant teams
105
.
What is needed is a system and method that can overcome the limitations of conventional systems by providing advanced teaming functionality which is capable of supporting both, similar and dissimilar network adapters, to improve the utility and performance of fault-tolerant adapter teams.
SUMMARY
Systems and methods consistent with the principles of the present invention address the need identified above and overcomes the prior art limitations by presenting a system and method that provides advanced teaming functionality which is capable of supporting both similar and dissimilar network adapters, to improve the utility and performance of fault-tolerant adapter teams.
Accordingly, a system and method, consistent with the principles of the present invention as embodied and broadly described herein, includes a team of multiple network adapters such that the team is capable of concurrently supporting network adapters of similar and dissimilar types. The capabilities of each of the network adapters in the team a
Bhat Gangadhar D.
Brown Daniel A.
Tran Han N.
Cangialosi Salvatore
Intel Corporation
LandOfFree
Method and system of providing advanced teaming... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and system of providing advanced teaming..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system of providing advanced teaming... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3147983