WO2016068968A1

WO2016068968A1 - Alternate address network login

Info

Publication number: WO2016068968A1
Application number: PCT/US2014/063302
Authority: WO
Inventors: Siamack Ayandeh
Original assignee: Hewlett Packard Enterprise Development Lp
Priority date: 2014-10-31
Filing date: 2014-10-31
Publication date: 2016-05-06

Abstract

Examples relate to alternate address network login. An example method includes receiving a login request to a fabric switch from a host on a network. Connection parameters are returned from the fabric switch in response to the login request. The returned connection parameters include; a first fabric address assignment having a first link to the fabric switch, a second fabric address assignment having an alternate link to the fabric switch, and an alternate local port name to the fabric switch. The example method includes detecting a link event on the fabric switch and applying a forwarding rule based on a type of link event detected to determine use of the first or the second fabric address.

Description

Alternate Address Network Login

Background

[0001] Computing networks can include multiple network devices including network devices such as routers, switches, hubs, and computing devices such as servers, desktop PCs, laptops, workstations, mobile devices and peripheral devices, e.g., printers, facsimile devices, and scanners, networked together across wired and/or wireless local and/or wide area network (LANs/WANs).

[0002] High Availability (HA) is a mode of operation for networks where redundant components are used to reduce service outage. The cost of high availability in fibre channel (FC)/fibre channel over Ethernet (FCoE) networks, measured in terms of the number of nodes, increases linearly with the number of switches in an HA cluster.

[0003] Currently a gateway proxy may be used to connect multiple hosts or virtual machines to a network fabric. The gateway proxy logs in to a switch on the network fabric on each link that it has with the switch to the network fabric. Any attempt by a host to connect to the network fabric is handled by the gateway proxy to register the host with the switch and receive an address for the host in the network fabric.

[0004] Multiple login attempts from the hosts are distributed amongst the links. Sometimes an active connection is disconnected and a host or virtual machine is forced to log back into the network in the face of link up/down events. This can negatively impact service and may affect a relatively large number of connections.

Brief Description of the Drawings

[0005] Figure 1 illustrates an example of a system including a switch in a network fabric connected to a plurality of hosts via a plurality of links consistent with the present disclosure. [0006] Figure 2A illustrates an example of a fibre channel over Ethernet (FCoE) forwarder (FCF) node according to the present disclosure.

[0007] Figure 2B illustrates an example of a gateway proxy switch according to an example of the present disclosure.

[0008] Figure 3 illustrates the structure of the addresses and the address components used by controlling FCFs to assign addresses according to the present disclosure.

[0009] Figure 4 illustrates an example of a login response control frame that can be returned by the logic and instructions of a switch in a network fabric according to one example of the present disclosure.

[0010] Figures 5A-5E illustrate an example of frame address rewrites consistent with at least one example of the present disclosure.

[0011] Figure 6 illustrates an affected connections control frame as may be generated, exchanged and processed by FCFs of a gateway proxy and/or a fabric switch, consistent with the present disclosure.

[0012] Figure 7A illustrates a connection list request for migration to a new link control frame consistent with the present disclosure.

[0013] Figure 7B illustrates a connection list reply for migration to a new link control frame consistent with the present disclosure.

[0014] Figure 8 illustrates an example method for alternate address login consistent with the present disclosure.

Detailed Description

[0015] Examples of the present disclosure may include network devices, systems, and methods, including executable instructions and/or logic thereon for alternate address network login. In some systems, a network gateway switching device, e.g., an N_Port Virtualizer (NPV) acts as a proxy between a host adapter and a switch in a network fabric. The proxy, e.g., NPV or gateway switch, may log in to a switch in the network fabric on each link that the proxy has with the switch in the network fabric.

[0016] For example, the proxy may log in using a fibre channel initialization protocol (FIP) login (FLOGI) process in fibre channel (FC) or fibre channel over Ethernet (FCoE) on links to the switch in the network fabric. Any login attempt from multiple hosts, e.g., host 1 , host 2, . . ., host N, is then connected by the proxy so that the host can interact with the switch to register a new device on the network and in the process communicate the host name and characteristics to the switch. In return, the proxy receives an address for that device on the host and the switch characteristics. This information is relayed back to the proxy and by the proxy back to the host as a login response. The address will be unique in the fabric with the fabric including numerous switch elements.

[0017] Multiple login attempts from the multiple hosts are distributed amongst multiple links between the gateway proxy and the switch on the network fabric. If a link is lost from service for one reason or another, then all the connections established with that link are lost and the associated hosts have to log back in to the switch on another link. Other scenarios, such as the addition of a new link may also trigger a re-distribution of the connections between the links, causing service disruption.

[0018] The procedure is relatively lengthy in time and service disruptive. The duration of a forced log-out and login may be of the order of seconds during which no service is available. That is, for the service disruption's duration, no useful payload data traffic is carried from the host. Log-out and login events and associated outages are often used as a metric for service level agreements and are closely monitored as an indicator of quality of a given transport technology.

[0019] In some technologies such as FC/FCoE the login and logout out events are fabric wide and are processed by the entire fabric including associated end points of a connection, e.g., storage targets. Therefore the processing burden can be fabric wide.

[0020] To remedy such a loss or traffic and service disruption outage, a switch in the network fabric is provided with instructions and logic to return an alternate address network login (shadow login). The additional procedure allows the host to continue its operation without the loss of its logged in connection in the face of such link up/down events or need for re-distribution of connections over links to the switch. This process decouples the distribution of the connection load from the host login process, allowing the host to continue its operation in the face of re-distribution. Examples can be used in Fibre Channel over Ethernet (FCoE) systems including T1 1 -SW6 standards under

development.

[0021] One example method includes receiving a login request to a fabric switch from a host on a network. Connection parameters are returned to the host from the fabric switch in response to the login request. The returned connection parameters include; a first fabric address assignment having a first link to the fabric switch, a second fabric address assignment having an alternate link to the fabric switch, and an alternate local port name to the fabric switch. The example method includes detecting a link event on the fabric switch and applying a forwarding rule based on a type of link event detected to determine use of the first or the second fabric address.

[0022] Figure 1 illustrates an example of a system 100 including a switch in a network fabric connected to a plurality of hosts via a plurality of links consistent with the present disclosure. As shown in Figure 1 , a plurality of host, Host 1 , Host 2, . . ., Host N, (104-1 , 104-2, 104-N), may connect to one or more switches, 1 10-1 , . . ., 1 10-M, in a switch fabric 100 via a plurality of links 1 1 1 -1 , 1 1 1 -2, 1 1 1 -3 (L1 , L2, . . . L3), etc. In at least one example the plurality of hosts 104-1 , 104-2, . . ., 104-N, connect to the one or more switches 1 10-1 , 1 10-2, 1 10-M, via a gateway proxy 102. In some examples the gateway proxy 102 can include a gateway proxy such as an N_Port Virtualizer (NPV). Alternatively, the gateway proxy can be implemented as computer executable instructions executed by a processing resource and/or logic on a soft switch, in the form of an application specific integrated circuit (ASIC), or on a network interface card (NIC), such as network adaptor card 1 12 shown on Host 1 in Figure 1 . In at least one example the network adaptor card 1 12 can be an N_Port network interface card. According to example of Figure 1 , a switch, e.g., 1 10-1 , connected to a plurality of hosts, e.g., 104-1 , . . ., 104-N, via a plurality of links 1 1 1 -1 , . . ., 1 1 1 -3, can include a fibre channel switch, e.g., a fibre channel over Ethernet (FCoE) forwarder (FCF) node, described and illustrated with Figure 2 [0023] As shown in the example of Figure 1 , the plurality of hosts, 104-1 , . . ., 104-N, can each support a plurality of virtual machines VM1 and VM2, 106- 1 and 106-2, which can be connected to a network adaptor card 1 12 via a soft- switch 108. As used herein the term Softswitch (vSwitch) is intended to mean a computer executable instructions running on a hypervisor. As used herein the term virtual machine (VM), e.g., VM1 and VM2, is intended to mean a software (computer executable instructions) instantiation of a physical host. A host in intended to mean a compute component having memory resources, processing resources and instructions stored in the memory resources and executable by the processing resources.

[0024] While examples herein may be described in connection with fibre channel over Ethernet (FCoE) networks and switches in a network or switch fabric, examples are not so limited. Examples may be applicable to a wide variety of systems requiring a network login.

[0025] The hosts 104-1 , . . ., 104-N, gateway proxy 102 and switch 1 10-1 in switch fabric can form a network. For example a network can attach to a storage area network (SAN), e.g., a switching fabric 100, to form a SAN fabric. A switch fabric 100 is a collection of fibre channel (FC) switches and/or fibre channel over Ethernet (FCoE) forwarders (FCF) which share a fibre channel fabric mapping (FCF_MAP) and run fabric protocols over their "E" ports

(E_Ports) and/or virtual E_Ports (VE_Ports). E_Ports are a type of port for connecting switches. F_Ports are a type of port for connecting to hosts and targets. Figure 1 , illustrates a plurality of F_Ports 1 13-1 , 1 13-2, 1 13-3, connecting independently to each of the plurality of links 1 1 1 -1 , 1 1 1 -2, 1 1 1 -3 which connect to the gateway proxy 102.

[0026] The FCF_MAP, referenced above, is a unique identifier for the switch fabric 100. Again, while switches and switching fabric are used in the example of Figure 1 , examples are not limited to switches and switching fabric for network devices. As used herein the term "virtual" in reference to a component, e.g., domain, port, connection and/or link, is intended to mean a logical connection versus an express physical connection. The term "virtual machine" is intended to mean a software instantiation of a physical host, e.g., physical computing component.

[0027] As mentioned above, the gateway proxy 102 can be an N-Port Virtualizer (NPV) gateway switch consisting of logic (shown in Figure 2B) similar to an FCF node (shown in Figure 2A), but with minimal FC control stack

(discussed below). Examples can apply to both FC and FCoE networks. Nodes are connected with links to form the SAN or "fabric". A fabric is a collection of switches which share a database and a so termed FC_MAP.

[0028] According to examples, described further below, program instructions, as stored in a fibre channel control stack and executable by a processing resource (e.g., within a control plane of a given switch), can execute instructions to detect a link event and provide an alternate address to connect a host or virtual machine to connect with a switch in a network fabric. That is the program instructions, e.g. control plane instructions of an FCF, can execute to change an address configuration in logic, e.g., hardware in the form of an application specific integrated circuit (ASIC) to process an alternate address.

[0029] Figure 2A illustrates an example of an FCoE forwarder (FCF) node according to an example of the present disclosure. As shown in the example of Figure 2A, an FCF node 202 can include a processor 220, e.g. processing resource, coupled to a memory 222. The memory 222 may include a fibre channel control stack 224, e.g., control plane software (computer executable instructions or program instructions). The program instructions may be executed to perform acts in connection with the examples of the present disclosure, e.g., to configure hardware of the FCF node to employ an alternate address.

[0030] Figure 2B illustrates an example of a gateway proxy 204

according to an example of the present disclosure. As shown in the example of Figure 2B, a gateway proxy 204 can include a processor 220, e.g. processing resource, coupled to a memory 222. The memory 222 may include a reduced, e.g., minimal, fibre channel (FC) control stack 225, e.g., control plane software (computer executable instructions or program instructions). The program instructions may be executed to perform acts in connection with the examples of the present disclosure, e.g., to configure hardware of the gateway proxy to employ an alternate address.

[0031] The memory 222 associated with an FCF node 202 of Figure 2A or a gateway proxy 204 of Figure 2B, may further include an Ethernet control stack 226, and a control state 228, including instructions executable to track and monitor a connection state of a given host, e.g., host 1 12-1 , . . ., 1 12-N in Figure 1 . According to examples of the present disclosure, program instructions associated with the FCF 202 control stack 224 are executed by the processing resource 220 to detect a link event and provide and alternate address to route traffic without an interruption of service by assigning a first fabric address, a second fabric address and an alternate port name as part of a network login process.

[0032] As part of the login process the switch 1 10-1 in the network fabric provides this additional information for the host 104-1 or virtual machine VM1 , VM2, etc., which is attempting to connect with the fabric 100. A VM/host name is associated with the given address. The address is carried as part of a header in transmitted frames and is used to route frames to their destination. As shown next in Figure 3, an address can be made up of a number of bytes in length, e.g., 6, 4, 3 bytes, etc., and may have a structure which reflects the topology of the network. The address structure may be used by the forwarding logic of the switches, e.g. fabric switch 1 10-1 and/or gateway proxy switch 102, to efficiently maintain table size and structure within such switch nodes which can be an expensive and limiting component to switches and switch nodes.

[0033] Figure 3 illustrates an example structure of an address and address components which may be used consistent with the present disclosure. The example of Figure 3, can be an FCoE example showing an address with three fields. The address 301 is shown including a domain ID portion 330, an area ID portion 332, and a port ID portion 334.

[0034] In this example, the domain ID port 330 can represent a switch in a switch fabric, e.g., switch 1 10-1 among the multiple switches in the switch fabric 100 in Figure 1 . The area ID portion 332 can represent a port of the switch, e.g., 1 13-1 , 1 13-2 or 1 13-3 in Figure 1 , which can be a physical or virtual link, connection with some transmission and switching capacity. The port ID portion 334 can uniquely identify a host or virtual machine, e.g., 104-1 , or VM1 , M2 (106-1 , 106-2), etc., that attempted to login over that port.

[0035] Hence, using this above example in reference to Figure 1 , there can be two fabric login addresses assigned to VM1 and VM2, respectively. These two fabric login address could be as follows: for VM1 : 10.1 .1 , for switch 10 (1 10-1 ), link 1 (L1 ), and host/VM1 ; and for VM2: 10.2.2, for switch 10 (1 10- 1 ), link 2, and host/VM2.

[0036] In a previous network operating environment, if service on link 1 (L1 ) were disrupted (e.g., via a link failure, etc.), VM1 would need to log out of the switch fabric 100 i.e., switch 10 (1 10-1 ) in this example, and log back in on a different link, e.g., link 3 (L3). As such, VM1 would receive a "new" fabric login address.10.3.5, where "5" to the port ID portion represents the new unique host ID or VM ID, e.g. VM1 , that is attempting to log back in.

[0037] In contrast, examples of the present disclosure include logic and/or instructions that are executed by the switch, e.g., switch 10 (1 10-1 ) to return connection parameters from the switch to the host 104-1 in response to a login request. However, in the present disclosure, the returned connection parameters include both a first fabric address assignment having a first link to the fabric switch, a second fabric address assignment having an alternate link to the fabric switch. Additionally, an alternate local port name to the fabric switch is returned.

Figure 4 illustrates an example of a login response control frame that can be returned by the logic and instructions of a switch in a network fabric (1 10-1 in Figure 1 ) according to one example of the present disclosure. As shown in Figure 4, the returned login response control frame can include a field 402 having a set of normal login response parameters for the network. In field 404, a first fabric address can be provided to the host (104-1 in Figure 1 ) or virtual machine VM1 . In the example above, this first fabric address, or fabric login address, can be 10.1 .1 in the respective domain ID portion 330, area ID portion 332, and port ID portion 334 of the address structure shown in Figure 3. The normal address assignment, e.g., 10.1 .1 for VMI , is returned to VM1 and is used to identify it throughout the fabric (100 in Figure 1 ).

In field 406, a second fabric address assignment (or "alternate fabric address") having an alternate link to the fabric switch (1 10-1 in Figure 1 ) can be provided to the gateway proxy (102 in Figure 1 ). And, in field 408, an alternate local port name on the fabric switch (104-1 ) can be provided to the gateway proxy 102. As used in this disclosure the second fabric address may be referred to as a "shadow" login address or alternate address. The second fabric address can be assigned on an alternate port to the fabric switch (1 10-1 ). In the example above, the second fabric address (alternate address) assignment, can be 10.3.5 in the respective domain ID portion 330, area ID portion 332, and port ID portion 334 of the address structure shown in Figure 3. For example, this alternate address assignment can be on an alternate port (e.g., 1 13-3) on the switch (1 10-1 ) on an alternate link (L3 1 1 1 -3 in Figure 1 ) which acts as a standby switch port and link for that connection. Here, the port ID portion "5" (334 in Figure 3) again uniquely identifies the host (104-1 ) or virtual machine VM1 . The name and identity of the local port attached to the alternate link (L3) can enable a re-route around a link, e.g., link (L1 ), whose service may have been disrupted, e.g., due to a link failure.

[0038] In at least one example, the normal login response parameters 402 first fabric address 404, alternate fabric address 406, and local port name associated with the alternate address 408 are provided to a gateway proxy (102 in Figure 1 ) in the form of a login response control frame. In an alternate example, the normal login response parameters 402 first fabric address 404, alternate fabric address 406, and local port name associated with the alternate address 408 are provided to a gateway proxy (102 in Figure 1 ) in a fabric login response using an additional control field in such a way that allows an encapsulation with an additional header field for subsequent data frames.

[0039] The above additional information, described in connection with Figure 4 as an extended login response control frame, can be packaged and used as frame headers as an "alternate address" and can be carried and exchanged as a type, length, value (TLV) format used with control protocol exchanges. This additional information can be removed by a Softswitch (108 in Figure 1 ), network adapter card (1 12), or by the gateway proxy 102, using logic or instructions described in connection with Figures 2A and 2B, to allow for the re-transmission of an original login response to the host (104-1 ) or virtual machine VM1 .

[0040] Several following scenarios are given to illustrate examples of link down, link up, and/or link addition events by which this disclosure provides improved mechanisms for traffic re-distribution and to better balance traffic load in network systems. In some examples, logic and/or instructions to an FCF in the fabric switch (1 10-1 ) detect a link event on the fabric switch and apply a forwarding rule based on a type of link event detected to determine use of the first or the second fabric address for routing traffic. Additionally, in some examples, logic and/or instructions to the gateway proxy (102) detect a link event with the fabric switch (1 10-1 ) and apply a forwarding rule on the gateway proxy (102) based on a type of link event detected to determine use of the first or the second fabric address for routing traffic. In some examples, logic and/or instructions in both an FCF in the gateway proxy (102) and an FCF in the fabric switch (1 10-1 ) detect a link event with the fabric switch (1 10-1 ) and apply a respective forwarding rule based on a type of link event detected to determine use of the first or the second fabric address for routing traffic. The following examples call out a given actor in the network as either a gateway proxy (102) or a fabric switch (1 10-1 ).

[0041] Figures 5A-5F illustrate an example of frame address rewrites consistent with the present disclosure. The example in Figures 5A-5F is consistent with a "link down" event, but may also be used for discussion and description of a "link up" or a "link addition" event, described below. In one example link down event, logic and/or instructions on the gateway proxy (102) can detect a link down event, e.g., a failed link such as L1 used in the first fabric address of the above example. An example of detecting a link down event by the gateway proxy can include detecting a link failure or taking down a port for administrative reasons, etc. [0042] The gateway proxy would then use the logic and/or instructions, e.g., shown in Figure 2B and similar to the FCF of Figure 2A, to activate an alternate forwarding rule. One example of an alternate forwarding rule includes selecting the second fabric address to route traffic from the host 104-1 or virtual machine VM1 . As described in more detail herein, the logic and/or instructions may be executed to check the connections associated with the failed link, e.g., L1 . The connections associated with the failed link may be tracked and maintained in a table by instructions executing on the gateway proxy and may be stored and accessed from memory.

[0043] The gateway proxy 102 may further use logic and/or instructions to implement a selected alternate forwarding rule to detect incoming flows and/or frames affected by the failed link L1 as determined from checking the associated connections list stored in the table, such as shown and described in connection with Figure 6.

[0044] Figures 5A-5F illustrate an address mapping in reference to the example of Figure 1 and in reference to the same address examples given above in connection with Figure 4. Figure 5A illustrates an example of an incoming frame to the gateway proxy 102 from a host 104-1 or virtual machine VM1 . This frame can include a checksum value 502, data (payload) 504, and a header including a source identifier (address) 506 and a destination identifier (address) 508. Using the example from Figure 4, the source address 506 may be given as 10.1 .1 and the destination address 508 may be given as X.1 to an example switch "X" in the network fabric 100.

[0045] As shown in frame 5B, the logic in the gateway proxy 102 can rewrite the source address (506 in frame 5A) for incoming frames from the host 104-1 or virtual machine VM1 with the alternate source address 507, e.g., 10.3.5, and forward the affected incoming frame or flow on the alternate port towards the switch 1 10-1 in the network fabric 100.

[0046] Frame 5E illustrates an outgoing return frame from the switch 1 10- 1 in the network fabric 100 to the gateway proxy 102. As shown in Frame 5E, the source address 505 for a return frame from a switch "X" in the fabric may be given as X.1 . The destination address 51 1 in the return frame received to the gateway proxy 102 from switch 1 10-1 in the fabric 100 may be given as 10.3.5. As shown then in frame 5F, the logic in the gateway proxy can rewrite the destination address 509 for outgoing frames from the gateway proxy 102 towards the host 104-1 or virtual machine VM1 . This is a mapping from an alternate destination address provided by an FCF of the switch 1 10-1 to an original login address, e.g., a mapping of destination address from 10.3.5 in the example above to a destination address of 10.1 .1 .

[0047] Additionally, in this link down event example, logic and/or instructions on the switch 1 10-1 in the network fabric 100 can detect a link down event, e.g., a failed link such as L1 used in the first fabric address of the above example. As mentioned above, an example of detecting a link down event by the gateway proxy can include detecting a link failure or taking down a port for administrative reasons, etc.

[0048] The switch 1 10-1 would then use the logic and/or instructions, e.g., shown in the FCF of Figure 2, to activate an alternate forwarding rule. The switch 1 10-1 then may receive a frame such as frame 5B having a source address 507 given as 10.3.5. As shown in frame 5C, logic and/or instructions in the switch 1 10-1 can rewrite the source address 506 back to 10.1 .1 before sending the frame to switch X in the fabric 100 at destination address 508 X.1 . Hence, one example of an alternate forwarding rule in an FCF of switch 1 10-1 includes rewriting the source address for incoming frames from the host 104-1 or virtual machine VM1 with the original login fabric address of the host 104-1 or virtual machine VM1 and forwarding the frame towards the rest of the switch fabric 100 and towards the frames ultimate destination, e.g., towards switch X at destination address 509 X.1 .

[0049] Frame 5D shows a return, incoming frame to switch 1 10-1 in the fabric 100 from switch X. Frame 5D includes a source address 505 as X.1 and a destination address 509 as 10.1 .1 . The logic and/or instructions in an FCF of the switch 1 10-1 will be used to select the second fabric address to route the frame from the switch 1 10-1 to the host 104-1 or virtual machine VM1 . That is, the logic and/or instructions of the switch 1 10-1 can rewrite the destination address for outgoing frames towards the host 104-1 or virtual machine VM1 for traffic from the switch fabric 100. This is a mapping of the original fabric login address 10.1 .1 to the alternate fabric address 10.3.5. Hence frame 5E illustrates an outgoing frame from the switch 1 10-1 in the fabric 100 to the gateway proxy 102 having a rewritten destination address 51 1 as 10.3.5.

[0050] According to the present disclosure, a similar process can be employed for a "link up" event. As used herein a link up event is intended to mean an event where a failed link, e.g., failed link 1 (L1 ) in the example above, returns to service prior to the state of connections timing out.

[0051] In this example, logic and/or instructions on the gateway proxy 102 may detect a link up event analogous to the above described manner to detecting a link down event. The gateway proxy would then run a load balancing algorithm to detect affected connections and checking a table such as described in Figure 6. Next, the logic and/or instructions of the gateway proxy 102 can be used to send a control frame, as shown in Figure 6, to the switch 1 101 -1 in the fabric with a list of affected connections.

[0052] The logic and/or instructions may cause the gateway proxy to pause for a duration which allows a link buffer to clear for any frames

associated with the affected connections which may be in transit. This can be a function of a link transit speed. For example, in high speed links and

reasonable buffer sizes, the pause duration may only be a few milliseconds or less. The logic and/or instructions in the gateway proxy 102 would then activate an original forwarding rule, e.g., first fabric address assignment, for the identified connections. The original forwarding rule detects the incoming flows and frames affected.

[0053] As described above in connection with Figures 5A-5F, the logic and/or instructions of the gateway proxy then implement the original source address for incoming frames from the host or virtual machine, rather that the alternate address, e.g., second fabric address, and forward the traffic on the original port towards the switch 1 10-1 in the network fabric 100. On return, a frame from the switch 1 10-1 has its original destination address 509 which may be left alone by the gateway proxy for outgoing frames toward the host or virtual machine. Hence, here, the mapping is from an alternate address to an original fabric login address, e.g., 10.1 .1 .

[0054] Also in this example, logic and/or instructions on the switch 1 10-1 can cause the switch to detect the link up event and receive and process the control frame with the list of affected connections, e.g., from the gateway proxy. Here, the switch may pause for a duration which allows a link buffer to clear for any frames associated with the affected connections which may be in transit. Again, this is a function of link speed and, for high speed an accepted buffer sizes, the pause duration can be a few milliseconds or less.

[0055] The logic and/or instructions in the switch 1 10-1 would then activate the original forwarding rule, first fabric address, for the identified connections moving to back to original link, e.g., link 1 (L1 ) in the above example, and the alternate forwarding rule detects the incoming flows and frames affected.

[0056] As described above in connection with Figures 5A-5F, the logic and/or instructions of the switch not have to rewrites the source address for incoming frames from the host or virtual machine, now no longer having the alternate address (e.g., 10.3.5), and will forward toward the rest of the switch fabric toward the ultimate destination, e.g., destination address X.1 in the above example in frame 5C. On return, a frame from switch X to switch 1 10-1 will no longer need to have its destination address 51 1 rewritten by the switch 1 10-1 for outgoing frames towards the host or virtual machine, e.g. as shown in frame 5E. This is a mapping will already be the original fabric login address.

[0057] As described in more detail herein, the logic and/or instructions can be executed to check the connections associated with the failed link, e.g., L1 . The connections associated with the failed link may be tracked and maintained in a table by instructions executing on the gateway proxy and may be stored and accessed from memory.

[0058] Figure 6 illustrates an affected connections control frame as may be generated, exchanged and processed by logic and/or instruction of the FCFs in a gateway proxy 102 and/or in a switch 1 10-1 of the network fabric 100 shown in Figure 1 , consistent with the present disclosure. As shown in Figure 6, the affected connection list control frame can include a source switch name field 602, a destination switch name field 604, and a target local port name field 606. The affected connections control frame additionally contains a number of affected connections field 608 and a field for a local port name 610-1 , . . ., 610- N and address 612-1 , . . ., 612-N for each affected connection.

[0059] Detection of a local link failure is usually accomplished in very short order, micro seconds or milliseconds at most. While a link down event may cause loss of some frames, the detections and switch over to an alternate forwarding rule is as fast as current technology can respond to failures. Any lost or out of order frames are recovered by higher layer protocols. For example, in FC/FCoE the recovery is at the exchange layer.

[0060] A further example of the present disclosure includes a "link addition" event. A similar process to that which is described above, can be employed for a "link addition" event. As used herein a link addition event is intended to mean an event where a new link is added as a connection to a port on a fabric switch to a network.

[0061] In one link addition example, logic and/or instructions on the gateway proxy 102 may detect a link addition event analogous to the above described manner to detecting a link up/down event. The gateway proxy would then run a load balancing algorithm to detect affected connections and to create a table such as described in Figure 7A below. Next, the logic and/or

instructions of the gateway proxy 102 can be used to send a control frame to the switch 1 10-1 in the fabric with a list of connections to migrate. The gateway proxy can receive a response from the fabric switch 1 10-1 for migrating connections such as described in Figure 7B below.

[0062] The logic and/or instructions in the gateway proxy 102 would then activate an alternate forwarding rule such using an alternate fabric address and local port name as could be provided in the control frame shown in Figure 4, for the identified connections. The alternate forwarding rule detects the incoming flows and frames affected.

[0063] As described above in connection with Figures 5A-5F, the logic and/or instructions of the gateway proxy 102 then rewrites the source address for incoming frames from the host or virtual machine with an alternate address and forwards on an alternate port towards the switch 1 10-1 in the network fabric 100. On return, a frame from the switch 1 10-1 has a destination address rewritten by the gateway proxy for outgoing frames toward the host or virtual machine, e.g., as shown in frame 5F. This is a mapping from an alternate address to an original fabric login address.

[0064] Also in this example, logic and/or instructions on the switch 1 10-1 can cause the switch to detect the link addition event and receive, e.g., from the gateway proxy, and process the control frame with the list of connections to migrate to this new link as shown and described in connection with Figure 7A. The switch can respond to the control frame received in Figure 7A with a control frame such as shown in Figure 7B.

[0065] The logic and/or instructions in the switch 1 10-1 would then activate an alternate forwarding rule on the switch for the connections moving to the new link and the alternate forwarding rule will detect the incoming flows and frames affected.

[0066] As described above in connection with Figures 5A-5F, the logic and/or instructions of the switch then rewrites the source address for incoming frames from the host or virtual machine, having now an alternate address, with the original fabric login address of the host or virtual machine and forward toward the rest of the switch fabric toward the ultimate destination, e.g., destination address X.1 in the above example in frame 5C. On return, a frame from switch X to switch 1 10-1 has its destination address 51 1 rewritten by the switch 1 10-1 for outgoing frames towards the host or virtual machine, e.g. as shown in frame 5E. This is a mapping from of the original fabric login address to the alternate address.

[0067] As described in more detail herein, the logic and/or instructions can be executed to check the connections associated with an added link. The connections associated with an additional link may be tracked and maintained in a table by instructions executing on the gateway proxy and may be stored and accessed from memory. [0068] Figure 7A illustrates a connections list request for migration to a new link control frame as may be generated, exchanged and processed by logic and/or instructions of the gateway proxy 102 and/or in a switch 1 10-1 of the network fabric shown in Figure 1 . As shown in Figure 7A, the connections list request for migration to a new link control frame can include a source switch name field 702, a destination switch name field 704, and a target local port name field 706. The connections list request for migration to a new link control frame additionally contains a number of affected connections field 708 and a field for an address 710-1 , . . ., 710-N for each affected connection.

[0069] Figure 7B illustrates a connection list reply for migration to a new link control frame consistent with the present disclosure. The connection list reply for migration to a new link control frame as may be generated, exchanged and processed by logic and/or instructions of the gateway proxy 102 and/or in a switch 1 10-1 of the network fabric shown in Figure 1 . The connection list reply for migration to a new link control frame example shown in Figure 7B includes a source switch name field 702, a destination switch name field 704, and a target local port name field 706 as with the control frame in Figure 7A. The

connections list reply for migration to a new link control frame also contains a number of affected connections field 708 and additionally contains a field for an alternate address 71 1 -1 , . . ., 71 1 -N for each affected connection.

[0070] Figure 8 illustrates an example method for alternate address login consistent with the present disclosure. As shown at block 802 the method can include receiving a login request to a fabric switch from a host connected to a network. As shown and described in connection with Figure 1 , a plurality of ports can connect a fabric switch via a plurality of links to a plurality of hosts. The host may additionally be a virtual machine.

[0071] At block 804, connection parameters including a first fabric address can be returned to the host and/or virtual machine and a second fabric address can be returned to the gateway proxy. As described herein, logic and/or instructions, used to return the connection parameters, including the first and the second fabric address, can be through a gateway proxy, Softswitch or network adapter card connection from the fabric switch. In one example the connection parameters are returned in a fabric login response using an additional control field in such a way that allows an encapsulation with an additional header field or rewriting of the address in the header for subsequent data frames. In another example, the connections parameters are returned using a control frame separate from a fabric login response on the alternate link from a first link. Figure 4 illustrates an example of such a login response control frame.

[0072] At block 806, a link event may be detected on a gateway proxy, Softswitch, or network adapter card and a list of affected connections can be generated, exchanged and processed as a control frame between a gateway proxy and a fabric switch. The link event may be detected and its type determined by the gateway proxy and fabric switch hardware and software as referenced in Figures 2A and 2B.

[0073] Block 808 further illustrates the method includes applying a forwarding rule using logic and/or instructions on a gateway proxy, Softswitch, network adapter card and/or fabric switch, based on a type of link event detected, to select between a first and a second fabric address for routing traffic between the host or virtual machine and a fabric switch on the network fabric. One example of selecting between a first and a second fabric address is given in Figures 5A-5F.

[0074] Hence, examples disclosed herein are not service affecting. In the rare scenario where frames are lost due to a link down event or out of order due to a link up event, upper layer protocols can recover through retransmission. The techniques disclosed herein can thus improve the service level agreement for a transport technology which is often the cause of service calls and complaints about outages. And, link up/down events can be reduced to a local event between a gateway proxy, Softswitch or network adapter card and the first hop switch (fabric switch), as opposed to a fabric wide event and thus such events will not affect the associated end points.

[0075] In the detailed description of the present disclosure, reference is made to the accompanying drawings that form a part hereof, and in which is shown by way of illustration how examples of the disclosure may be practiced. These examples are described in sufficient detail to enable those of ordinary skill in the art to practice the examples of this disclosure, and it is to be understood that other examples may be utilized and that process, electrical, and/or structural changes may be made without departing from the scope of the present disclosure. As used herein, the designators "A", "X", "Y", "Z", "E", "C", "D", "M", "N," "M," "P", etc., particularly with respect to reference numerals in the drawings, indicate that a number of the particular feature so designated can be included with examples of the present disclosure. The designators can represent the same or different numbers of the particular features.

[0076] The figures herein follow a numbering convention in which the first digit or digits correspond to the drawing figure number and the remaining digits identify an element or component in the drawing. Similar elements or components between different figures may be identified by the use of similar digits. For example, 102 may reference element "02" in Figure 1 , and a similar element may be referenced as 202 Figure 2. Elements shown in the various figures herein can be added, exchanged, and/or eliminated so as to provide a number of additional examples of the present disclosure. In addition, the proportion and the relative scale of the elements provided in the figures are intended to illustrate the examples of the present disclosure, and should not be taken in a limiting sense.

[0077] Although specific examples have been illustrated and described herein, those of ordinary skill in the art will appreciate that an arrangement calculated to achieve the same results can be substituted for the specific examples shown. This disclosure is intended to cover adaptations or variations of one or more examples of the present disclosure. It is to be understood that the above description has been made in an illustrative fashion, and not a restrictive one. Combination of the above examples, and other examples not specifically described herein will be apparent to those of skill in the art upon reviewing the above description. The scope of the one or more examples of the present disclosure includes other applications in which the above structures and methods are used. Therefore, the scope of one or more examples of the present disclosure should be determined with reference to the appended claims, along with the full range of equivalents to which such claims are entitled.

Claims

What is claimed:

1 . A method for alternate address network login, comprising:

receiving a login request to a fabric switch from a host on a network; returning connection parameters to the host and a gateway proxy from the fabric switch in response to the login request, wherein the connection parameters include:

a first fabric address assignment having a first link to the fabric switch;

a second fabric address assignment having an alternate link to the fabric switch; and

an alternate local port name to the fabric switch;

detecting a link event on the fabric switch; and

applying a forwarding rule based on a type of link event detected to determine use of the first or the second fabric address.

2. The method of claim 1 , wherein returning connection parameters to the host, comprises:

returning the connection parameters in a fabric login response using an additional control field in such a way that allows an encapsulation with an additional header field or address rewrite for subsequent data frames.

3. The method of claim 1 , wherein returning the connection parameters to the host, comprises:

returning the connections parameters using a control frame separate from a fabric login response on the alternate link from the first link.

4. The method of claim 1 , wherein:

receiving the login request comprises receiving the login request from a virtual machine through a virtual gateway proxy between the host and the fabric switch;

additionally detecting the link event on the proxy; and applying the forwarding rule on both the fabric switch and the virtual gateway proxy.

5. The method of claim 1 , wherein returning the first fabric address assignment and the second fabric address assignment comprises:

returning the first fabric address assignment to a proxy for a virtual machine on the host to connect the virtual machine to the fabric switch on the first link;

returning the second fabric address assignment having the alternate local port on the alternate link as a standby switch port and link between the proxy and the switch fabric; and

returning the alternate local port name attached to the alternate link to allow a re-route of frames around the first link.

6. A fabric switch, comprising:

a plurality of ports to connect the fabric switch via a plurality of links to a plurality of hosts;

logic to, in response to receiving a network login request from a host: assign a login fabric address to the host and return the login fabric address to the host to connect frames from the host to the fabric switch on a first link;

assign an alternate fabric address to the host and return the alternate fabric address to a gateway proxy, the alternate fabric address being to an alternate switch port on an alternate link; and

return a local port name associated with the alternate fabric address to the gateway proxy.

7. The switch of claim 6, wherein the switch is a fibre channel over Ethernet forwarder (FCF) node comprising logic to:

detect a link event; and

select between routing frames using the login fabric address or the alternate fabric address based on the type of link event detected.

8. The switch of claim 6, wherein the switch is to connect to a gateway proxy, and the gateway proxy comprises logic to:

detect a link down event on the first link;

activate an alternate forwarding rule for connections associated with the first link;

detect frames received to the gateway proxy that are affected by the alternate forwarding rule;

rewrite a host source address for an incoming frame from the host to the gateway proxy with the alternate fabric address and forward the frame on the local port name associated with the alternate fabric address; and

rewrite a destination address for an outgoing frame toward the host from the switch fabric from the alternate fabric address to the host source address.

9. The switch of claim 6, wherein the switch comprises logic to:

detect a link down event on the first link;

detect frames received to the switch that are affected by the alternate forwarding rule;

rewrite a source address for an incoming frame from a gateway proxy to the switch with the login fabric address of the host and forward the frame toward a destination client in the switch fabric; and

rewrite a destination address for an outgoing frame toward the gateway proxy from a host address to the alternate fabric address.

10. A virtual gateway proxy to a network fabric, comprising:

a plurality of ports to connect the virtual gateway proxy via a plurality of links to a plurality of hosts to receive login requests from virtual machines (VMs) associated with the plurality of hosts; and

logic to: receive a login fabric address assigned to a virtual machine to connect frames from the virtual machine to a fabric switch on via a first link; receive an alternate fabric address assigned to the virtual machine on the host to connect frames from the virtual machine to the fabric switch via an alternate link to an alternate port on the fabric switch; and

detect a link event on the first link.

1 1 . The proxy of claim 10, wherein the virtual gateway proxy comprises logic to:

detect a link down event;

rewrite a VM source address for an incoming frame from the host to the virtual gateway proxy with the alternate fabric address and forward the frame to the alternate switch port associated with the alternate fabric address; and

rewrite a destination address for an outgoing frame toward the host from the fabric switch from the alternate fabric address to the host source address.

12. The proxy of claim 10, wherein the virtual gateway proxy is to connect to the fabric switch via a plurality of links to a plurality of ports on the fabric switch, and the fabric switch comprises logic to:

rewrite a source address for an incoming frame from the virtual gateway proxy to the fabric switch with the login fabric address of the host and forward the frame toward a destination client in the network fabric; and

rewrite a destination address for an outgoing frame toward the virtual gateway proxy from a host address to the alternate fabric address.

13. The proxy of claim 10, wherein the virtual gateway proxy includes logic to:

detect a link up event for the first link;

determine affected connections by checking an affected connections list; send a control frame to a fabric switch with a list of affected connections; pause a duration to allow a link buffer to clear any frames associated with the affected connections; and

restore an original forwarding rule.

14. The proxy of claim 10, wherein the virtual gateway proxy is to connect to the fabric switch via a plurality of links to a plurality of ports on the fabric switch, and the fabric switch comprises logic to:

detect a link up event for the first link;

receive a control frame with a list of affected connections;

pause a duration to allow a link buffer to clear any frames associated with the affected connections; and

restore an original forwarding rule.

15. The proxy of claim 14, wherein:

both the virtual gateway proxy and the fabric switch include logic to detect a new link entering service;

the virtual gateway proxy includes logic to:

run a load balancing algorithm to detect affected connections; and send a control frame from the virtual gateway proxy to the fabric switch with a list of connections to migrate;

the fabric switch includes logic to receive and process the control frame; and both the virtual gateway proxy and the fabric switch include logic to apply an alternate forwarding rule in response to the new link entering service.