JP6471066B2 - Network management apparatus and address setting method - Google Patents

Description

  The present invention relates to address setting of a network device in a network including a plurality of CLOS networks.

  In recent years, communication networks have become larger and more complex. That is, the number of network devices has increased, and their connection relations have become complicated. For this reason, the setting and management of network devices have become complicated.

  Therefore, for example, network device settings are automated by SDN (Software Defined Networking) technology. As an SDN, for example, in OpenContrail (Non-patent Document 1) published as an open source, a virtual network can be freely and semiautomatically set according to an instruction from a user. Further, Non-Patent Document 2 discloses specific contents of this OpenContrail.

"OpenContrail-An open-source network virtualization platform for the cloud.", [Online], OpenContrail, [searched July 30, 2015], Internet <URL: http://www.opencontrail.org/> “CONTRAIL Architecture”, [online], JUNIPER NETWORKS, [Search June 22, 2015], Internet <URL: http://www.juniper.net/jp/jp/local/pdf/whitepapers/2000535-jp. pdf>

  However, in the above-described prior art, the automation of the setting of the network device is targeted for the overlay network portion. In other words, assuming that there is an environment in which a physical network device is already configured and communication is ensured, settings for configuring a virtual network thereon are performed using software.

  On the other hand, in recent years, a network including a plurality of CLOS networks having a plurality of Spine devices and a plurality of Leaf devices in a so-called Spine & Leaf type two-layer configuration is often used from the viewpoint of high scalability and the like. In such a network, there has never been a technique for automatically setting the address of a physical network device.

  Therefore, an object of the present invention is to automatically set an address of a physical network device in a network including a plurality of CLOS networks.

In order to solve the above problems, the present invention is a network management device for setting addresses of the Spine device and the Leaf device in a network including a plurality of CLOS networks having a plurality of Spine devices and a plurality of Leaf devices. A starting point address that is a starting point of addresses to be set in the Spine device and the Leaf device, a storage unit that stores the number of the Spine devices and the number of Leaf devices for each CLOS network, and the starting point address, and Based on the number of the Spine devices and the number of Leaf devices for each CLOS network, a start address for each CLOS network that is a start point of the address for each CLOS network is determined, and for each CLOS network, the start point for each CLOS network Address, number of Spine devices and Le a processing unit for setting the address of the Spine device and the Leaf device based on the number of af devices, the processing unit being a product of the number of the Spine devices and the number of the Leaf devices for each CLOS network. The cluster ID of the CLOS network is determined based on the descending order of the CLOS network, the starting address for each CLOS network is determined so that the CLOS network is arranged in order from the descending order of descending order, and the Spine device of the one CLOS network The address of the Leaf device is set after the network-specific start address of the CLOS network and before the network-specific start address of the CLOS network of the next cluster ID .

  Accordingly, the addresses of the Spine device and the Leaf device can be automatically set only by registering the start point address, the number of Spine devices and the number of Leaf devices for each CLOS network.

  Further, the present invention is a CLOS network management device installed for each CLOS network managed by the network management device, wherein either the Spine device or the Leaf device is added in the CLOS network of its own. The address of the added device is determined based on the number of Spine devices and the number of Leaf devices in the CLOS network of itself, and the determined address is notified to the added device. Of the existing Spine device and Leaf device, the setting information for communication with the added device is notified to the device connected to the added device.

  As a result, when adding a Spine device or Leaf device, the address of the device to be added is determined, the determined address is notified to the added device, and added to the existing Spine device or Leaf device. The setting information for communication with the device to be notified can be notified.

  Further, according to the present invention, the storage unit stores a CLOS network start point interface address that is a start point for determining an interface address for each CLOS network used for connection between the CLOS networks. The interface address for each CLOS network is determined based on the interface address between the CLOS networks, the number of the Spine devices and the number of the Leaf devices for each CLOS network, and the determined interface address for each CLOS network is determined. , By transmitting to each CLOS network management device installed for each CLOS network and causing the CLOS network management device to notify the subordinate Spine device and Leaf device. And performing connection between CLOS network.

  Thereby, connection between CLOS networks can be performed automatically.

  According to the present invention, the address of a physical network device can be automatically set in a network including a plurality of CLOS networks.

It is a figure which shows the structural example of the network of this embodiment. It is explanatory drawing about communication etc. in a cluster. It is a figure which shows the example of an address calculation result about one cluster. It is explanatory drawing regarding prior information registration. It is a sequence diagram of the process which calculates and distributes an address. It is a structural example of each information used by the process of FIG. 5A. (A) is a sequence diagram of processing for generating a connection information table and a logical IF information table. (B) is a figure which shows the example of an address calculation result about one cluster. (A) is a structural example of a logical IF information table. (B) is a structural example of a connection information table. It is a sequence diagram of a process for adding network devices. It is a figure which shows the structural example of one cluster and a relay apparatus. (A) is a structural example of a logical IF information table. (B) is a structural example of a connection information table. It is a sequence diagram of the process which connects between clusters.

  Hereinafter, modes for carrying out the present invention will be described. In the present embodiment, a network management device that performs address setting of a Spine device and a Leaf device in a network including a plurality of CLOS networks including a plurality of Spine devices and a plurality of Leaf devices will be described. In recent years, CLOS networks are often used in data centers and the like, and communicate using, for example, ECMP (Equal Cost Multi Path; see “https://tools.ietf.org/html/rfc2992”).

  FIG. 1 is a diagram illustrating a configuration example of a network S according to the present embodiment. The network S includes a host controller 1 (network management device), a backbone network 2, and a plurality of clusters 3 (CLOS network).

The host controller 1 includes a processing unit 11, a storage unit 12, and the like, and manages and controls the entire network S (details will be described later).
The backbone network 2 includes a relay device 21 that relays communications such as a plurality of transmission devices and layer 2 switches. The relay device 21 mediates communication between spine switches 32 (Spine devices) of two different clusters 3.

  The cluster 3 is a CLOS network having a so-called Spine & Leaf type two-layer configuration and having a plurality of Spine devices (first layer devices) and a plurality of Leaf devices (second layer devices). In the example of FIG. The four clusters are provided. Hereinafter, the cluster 3 may be referred to as “Cluster1 to 4”, “Cluster1”, or the like.

  Of Cluster 3, Cluster 1 will be described as a representative. Cluster 1 includes a controller 31 (CLOS network management device), a plurality of spine switches 32 (Spine 1 and 2) (network devices), and a plurality of leaf switches 33 (Leaf 1 to 4). (Leaf device, network device), a plurality of servers 34 are provided. Note that the number of spine switches 32 and the number of leaf switches 33 in each cluster 3 in FIG. 1 are examples, and are not limited to these, and may be more or less than this example.

  The controller 31 is connected to all the leaf switches 33. The plurality of spine switches 32 and the plurality of leaf switches 33 are connected by a full mesh. Each of the plurality of servers 34 is connected to one of the leaf switches 33.

  The route advertisement, M (Management) plane, and U (User) plane are as illustrated.

  Next, communication and the like in the cluster will be described with reference to FIG. The controller 31 of the cluster 3 (Cluster 1 as representative in FIG. 2) includes a DB (Data Base) 311, a setting control unit 312, a DHCP (Dynamic Host Configuration Protocol) server unit 313, and a Syslog server unit 314. Note that the DHCP server unit 313 is physically included in the controller 31 here, but is not limited thereto, and may be physically separated to the outside. The same applies to the DB 311.

  A controller 31 including a DB 311 and a DHCP server unit 313 exists for each cluster 3 and communicates with the host controller 1 with IP connectivity that can be connected using an IP address.

  Further, independent of the connection between the spine switch 32 and the leaf switch 33 (connection through which data traffic circulates), the spine switch 32 and the leaf switch 33 are each connected to the controller 31 (in the example of FIG. Through).

The DHCP server unit 313 exists in the same layer 2 segment as the spine switch 32 and the leaf switch 33 in order to receive DHCP data.
In the present embodiment, the spine switch 32 and the leaf switch 33 are described as L3 switches as an example, but may be realized as host nodes such as routers, PCs, and servers having equivalent functions.

  In the present embodiment, a mechanism will be described in which when the leaf switch 33 is added, the host controller 1 automatically calculates the address and notifies the controller 31 of the address.

First, parameters to be used will be described.
-Interface address start point IP address: X (input value)
Loopback address source IP address: Y (. When there is no input value input value, according to equation (4))
Management address start point IP address: Z (input value. If there is no input value, use equation (6))
The “start point” in each start point IP address means the start point of the calculation.

Cluster ID: C (start value: 1)
(C is the descending order of the product of the maximum number of spines and maximum number of leaves)
In the example of FIG. 1, the products of the maximum number of spines and the maximum number of leafs of Cluster 1 to 4 are 2 × 4 = 8, 2 × 5 = 10, 2 × 20 = 40, and 2 × 3 = 6, respectively. Therefore, the descending order of the products of Cluster 1 to 4 is 3, 2, 1, and 4.
Accordingly, the cluster IDs of Cluster 1 to 4 are “3”, “2”, “1”, and “4”, respectively.

・ Maximum number of Spine: Smax _C
・ Maximum number of Leafs: Lmax _C
・ Maximum cluster ID: Cmax

  Here, the starting IP addresses of the interface address, loopback address, and management address are used, but the essential information is the interface address starting IP address. The host controller 1 holds this together with other information (such as the above parameters) in the host DB of the storage unit 12.

（但し、Ｃ＞１の場合のみ。Ｃ＝１の場合Ｕ_Ｃ＝０）。
なお、log₂(Smax_i×Lmax_i)の両側の記号は、天井関数を表す。 Then, the host controller 1 performs the following calculation.

(However, only when C> 1, U _C = 0 when _C = 1).
The symbols on both sides of log ₂ (Smax _i × Lmax _i ) represent the ceiling function.

（但し、Ｃ＞１の場合のみ。Ｃ＝１の場合Ｖ_Ｃ＝０）

(However, only when C> 1, V _C = 0 when _C = 1)

X _C = X _{+ U C} ··· (3)
Y = X _{Cmax + 1} Formula (4)
Y _C = Y + V _C Formula (5)
Z = Y _{Cmax + 1} Expression (6)
Z _C = Z + V _C (7)
However, Smax ₀ = 0, Lmax ₀ = 0

Here, paying attention to the cluster 3 whose cluster ID is “C”, the parameters used are as follows.
-Interface address start point IP address: X _C (Start point address by CLOS network)
-Loopback address start point IP address: Y _C
Management address start point IP address: Z _C
・ Maximum number of Spine: Smax _C
-Maximum number of Leafs: Lmax _C
・ Spine device number (start value: 1): S _n (n = 1, 2, 3,..., Smax _C )
Leaf device number (start value: 1): L _k (k = 1, 2, 3,..., Lmax _C )

Then, the Sn-Lk link Sn side IP address is calculated by the following equation (8).
_{X C + ((Sn-1} ) * Lmax C + (Lk-1)) * 2 ··· formula (8)

The Sn-Lk link Lk side IP address is calculated by the following equation (9).
X _C + ((Sn−1) * Lmax _C + (Lk−1)) * 2 + 1 Expression (9)

Also, each address is calculated as in the following formulas (10) to (13).
Sn loopback IP address = Y _C + S _n Expression (10)
Sn management IP address = Z _C + S _n Expression (11)
Lk loopback IP address = Y _C + Smax _C + L _k Expression (12)
Lk management IP address = Z _C + Smax _C + L _k Expression (13)

An example of the calculation result is shown in FIG. In FIG. 3, X _C = 10.0.0.0 is assumed for Cluster1 in FIG. As the idea of number assignment according to equations (8) and (9), an even address is assigned from the first port of the first device number in the ascending order of the Spine device number, and the leaf device opposite to the port of each Spine device is assigned. The address of +1 is assigned to the port.

  For example, in Spine1, the interface address “10.0.0.0 (indicated as“ .0 ”in FIG. 3; the same applies hereinafter)” is assigned to the port “1”, and the interface address “10.0.0.2” is assigned to the port “2”. Has been granted. In Leaf1, the interface address “10.0.0.1” is assigned to the port “1”. In Leaf 2, the interface address “10.0.0.3” is assigned to the port “1”.

FIG. 4 and subsequent figures show a processing sequence, examples of data held by each controller, and the like.
FIG. 4 is an explanatory diagram regarding prior information registration.
First, when the host controller administrator 1M registers the switch model information 5 (FIG. 4B) for identifying the network device used in each cluster 3 by using the host controller 1 (step S1 in FIG. 4A). The host DB (FIG. 2) in the storage unit 12 of the host controller 1 is updated (step S2).

  In the switch model information 5 of the example shown in FIG. 4B, a message that is supposed to be transmitted last when the apparatus is activated is registered as an activation completion determination message corresponding to the model ID and model name. Also, set the startup image file path and startup configuration file if possible.

  Next, when the host controller administrator 1M uses the host controller 1 to instruct distribution of the switch model information 5 (step S3), the host controller 1 distributes the switch model information 5 to each controller 31 (step S4).

  Next, when the controller manager 31M registers intra-cluster device information using the controller 31 (step S5), the controller 31 updates the DB 311 (step S6). Specifically, in the intra-cluster device information table 6 shown in FIG. 4C, the device type ID, device number, model ID, MAC (Media Access Control) address (for device individual identification), flag (present in the network) Each information of “1” when registered, “0” when not present) is registered.

Next, an address distribution method will be described with reference to FIGS. 5A and 5B.
First, when the upper controller administrator 1M registers the WAN information 7 (FIG. 5B (a)) using the upper controller 1 (step S11 in FIG. 5A), the upper DB (FIG. 2) of the upper controller 1 is updated. (Step S12).

  In the WAN information 7 in the example shown in FIG. 5B (a), the interface address start point IP address (start point IF (Interface) address: a value corresponding to “X”), the loopback address start point IP address (start point Lo address: “Y”) ), Management address start point IP address (start point management address: value corresponding to “Z”), and inter-cluster start point IF address (inter-CLOS network start point interface address) are specified as input values.

  Next, when the host controller administrator 1M registers the cluster information 8 (FIG. 5B (b)) using the host controller 1 (step S13), the DB 311 of the controller 31 is updated (step S14).

  The cluster information 8 in the example shown in FIG. 5B (b) includes items of cluster ID, cluster maximum spine number, cluster maximum leaf number, and controller address. Such cluster information 8 is stored in the DB 311 of the controller 31.

  Next, when the host controller manager 1M uses the host controller 1 to instruct the calculation of the cluster start point address and the like (step S15), the host controller 1 performs the calculation and distributes the calculation result to each controller 31 (step S15). S16).

  FIG. 5B (c) is cluster information 8a related to Cluster1 in the example as a calculation result, and is stored in the controller 31 of Cluster1. The cluster information 8a includes information on the maximum cluster spine number, the maximum cluster leaf number, the start point IF address (start point address for each CLOS network), the start point Lo address, and the start point management address. The same applies to the cluster information 8b, 8c, and 8d in (d), (e), and (f) of FIG. 5B.

  Next, creation of a connection information table and a logical IF information table will be described with reference to FIGS. 6A and 6B. When the host controller 1 distributes the calculation result to each controller 31 (step S21 in FIG. 6A (corresponding to step S16 in FIG. 5)), the controller 31 that obtained the cluster information (FIGS. 5C to 5F). Automatically creates the connection information table 92 in FIG. 6B (b) and the logical IF information table 91 in FIG. 6B (a) in the DB 311 (steps S22 and S23).

  In the logical IF information table 91 in FIG. 6B (a), the logical IF ID, device type ID, device number, and IP address (interface IP address) associated with the physical IF are as shown in FIG. 6A (starting point). The values are based on Cluster1) with an address of 10.0.0.160, a spine number of 2, and a leaf number of 4.

  In the logical IF information table 91, logical IFs are assigned in ascending order of physical IF numbers of each device (Spine device, Leaf device). As a method of acquiring the physical IF numbers in ascending order, for example, a method of acquiring ifIndex by SNMP (Simple Network Management Protocol) MIB (Management Information Base) and using it in ascending order is conceivable. Depending on the use environment, the physical IF does not always start from “1” as shown in FIG. 6B (a), and the value of the physical IF shown in FIG. 6B (a) is merely an example. For example, the rule for the name of the interface to be used (eg, Gigabit is prefixed like Gigabit0, Gigabit1,..., As in the case of the model ID “2” in the switch model information 5 in FIG. 4B). If it is determined, it is possible to obtain an ifName from the MIB, extract the ifName to the name, and use the ifIndex in ascending order.

  Further, unlike the notation in the logical IF information table 91 in FIG. 6B (a), in practice, a unique number for each cluster is registered in the physical IF.

  In the connection information table 92 in FIG. 6B (b), each of the device type ID, device number, physical IF, counter device type ID, counter device number, and counter physical IF is assumed on the assumption of the case shown in FIG. 6A (b). Information is stored.

  Next, with reference to FIG. 7, a method for adding devices using the above-described algorithm will be described. 7 assumes that the operations shown in FIGS. 4 to 6 have been performed. Here, it is assumed that Leaf 5 is added in Cluster 1 of FIG.

  First, the controller manager 31M registers device information using the controller 31, that is, updates the intra-cluster device information table 6 (step S301). Specifically, as shown in the intra-cluster device information table 6 of FIG. 7, Leaf5 device type ID, device number, model ID, and MAC address are registered in the bottom row.

  When the device information is registered, the DHCP setting in which the DHCP server unit 313 operates is updated (step S302), and the MAC address (or management address) is associated with the IP address (calculated by Expression (9)). Note that the activation image file path or the activation configuration file in the switch model information 5 (FIG. 4) may be used.

Next, the maintenance person M performs connection and activation of the expansion target apparatus A (Leaf5) (step S303).
Then, the expansion target apparatus A (Leaf5) makes a DHCP request (step S304), and the DHCP server unit 313 in the controller 31 makes a DHCP response (step S305).

  Further, the management address of the controller 31 is assigned to the expansion target device A (Leaf5) as the setting of the syslog server unit 314 (step S306). As this method, for example, DHCP option7 (http://www.iana.org/assignments) /bootp-dhcp-parameters/bootp-dhcp-parameters.xhtml).

  After performing the setting by giving in step S306, the expansion target device A (Leaf5) sends a start notification (including a start completion determination message) to the controller 31 (step S307).

  Next, the controller 31 updates the flag of the intra-cluster device information table 6 (FIG. 4) in the DB 311, that is, rewrites from “0” to “1” to indicate that it already exists in the network ( Step S308).

After step S308, the following settings are made.
In step S309, the controller 31 sets an interface address as a full-scale setting for the expansion target apparatus A (Leaf5), and activates an IGP (Interior Gateway Protocol) for the interface.
In step S309, the expansion target device A (Leaf 5) sends a completion response to the controller 31.

In step S311, the controller 31 sets the interface address to the interface connected to the expansion target apparatus A (Leaf5) and the IGP (Interior Gateway) as settings for the existing target apparatus B (Spine1, 2). Protocol) Enable etc.
In step S <b> 311, the existing target device B (Spine1, 2) sends a completion response to the controller 31.

  Steps S309 to S312 can be realized, for example, by setting using CLI (Command Line Interface) or setting using the NetConf protocol. Then, by using the logical IF information table 91 (FIG. 6B (a)), the calculated IP address of each interface of the device can be given, and a loopback IP address can be given if necessary. .

  In step S313, the controller 31 responds (displays) the result to the controller manager 31M.

  With the above mechanism, in the network S, the interface IP address as the starting point, the number of devices (Spine devices, Leaf devices) are determined, and the connectivity within the cluster 3 is obtained only by having information for identifying individual devices. Can be established.

Next, a method for connecting the clusters 3 will be described.
Regarding the configuration of the cluster 3 shown in FIG. 8A (Cluster 1 in FIG. 1), the lower two lines in the logical IF information table 91 in FIG. And the information on the lower two lines in the connection information table 92 in FIG. 8B (b) are registered.

  This registration may be performed manually or automatically by including it in the initial information in steps S22 and S23 in the sequence described in FIG. 6A (a). In this way, the interface connected to the relay device 21 is associated with the logical interface that is logically managed.

Next, processing of each device will be described with reference to FIG. Here, a case where Cluster1 and Cluster2 are connected is taken as an example.
First, the host controller administrator 1M registers logical link information in the host controller 1 (step S401). The logical link information is information indicating connection between logical interfaces managed by the controller 31 of each cluster 3.

  In the example shown in FIG. 9, the logical interface number “17” of the cluster ID “1” (the number of internal connection interfaces is 2 (number of spines) × 4 (number of leafs) × 2 (both ends) = 16) The logical interface number “21” of the opposite cluster ID “2” (the number of the internal connection interface is 2 (number of spines) × 5 (number of leafs) × 2 (both ends) = 20) is connected.

  The host controller 1 holds the inter-cluster start point IF address in the WAN information 7 of FIG. 5B (a), that is, holds the start point of the address allocated to the inter-cluster interface (interface address for each CLOS network). ing. Thereby, the host controller 1 automatically determines an address by the following calculation and notifies the controller 31a of Cluster1 (step S402).

(Address determination concept)
Allocate addresses with the same system connection.
Links are allocated with / 31 (IPv4) and / 127 (IPv6).
(variable)
-Own cluster ID: A (beginning with “1”) [obtained from cluster information 8 (FIG. 5B (b))]
Opposite cluster ID: B (beginning with “1”) [obtained from cluster information 8 (FIG. 5B (b))]
Total number of clusters: Cmax [obtained from cluster information 8 (FIG. 5B (b)), taking the maximum value]
-Number of set units: W (starts with “0”) [Maximum value is obtained from cluster information 8 (FIG. 5B (b)), maximum value of maximum Spine within cluster]
-Start point address: I (continuous to the address range of the 0 system plane and continues with the 1 system plane and 2 system plane)

(a formula)
・ If (A <B)
{I + W × Cmax × (Cmax-1) + Formula (14)
}
else if (A> B)
{
I + W x Cmax x (Cmax-1) + formula (15)
}

ただし、ａ_０＝０、ａ_ｋ＝Ｃｍａｘ−ｋ

_{However, a 0 = 0, a k} = Cmax-k

ただし、ａ_０＝０、ａ_ｋ＝Ｃｍａｘ−ｋ

_{However, a 0 = 0, a k} = Cmax-k

  Note that the system here (set number W) indicates the number of layer 3 ground connections between specific clusters, and is calculated as 0 system for the first cluster and 1 system for the second cluster. To do.

  If there is an address (cluster number, logical IF number, Ipv4_addr / Ipv6_addr = calculation result of the host controller 1) notification from the host controller 1 to the Cluster1 controller 31a in step S402, the Cluster1 controller 31a sends an IP address to the corresponding logical interface. To update its own DB 311 (update the logical interfaces “17” and “18” in the logical IF information table 91a in FIG. 8B) (step S403), and the Cluster1 target device 3a (each Spine, each Leaf) The address is set to (Step S404).

  When receiving an address setting completion response from the Cluster1 target device 3a (step S405), the Cluster1 controller 31a sends a completion response to the host controller 1 (step S406).

  Steps S407 to S411 are the same as steps S402 to S406 except that the Cluster1 controller 31a and the Cluster1 target device 3a are changed to the Cluster2 controller 31b and the Cluster2 target device 3b.

  The host controller 1 updates the host DB of the storage unit 12 in response to the completion responses in steps S406 and S411.

  As described above, according to the host controller 1 and the like according to the present embodiment, the start point address (interface address start point IP address) and the number of spine devices and the number of leaf devices for each CLOS network are registered in the host controller 1. You can set the address of Spine device and Leaf device automatically just by setting.

  Therefore, when network devices (Spine devices, Leaf devices) are added or when a network device failure is replaced, if only the information that identifies the individual device is registered by the maintainer without entering detailed settings, it will automatically Settings can be entered into the network device to establish connectivity. In addition, since IP addresses used for establishing the connectivity can be aggregated, an increase in the path can be suppressed when the CLOS networks are connected to each other.

  In addition, when adding a Spine device or Leaf device, the address of the device to be added is determined, the determined address is notified to the added device, and added to the existing Spine device or Leaf device. The setting information for communication with the apparatus can be notified.

  Further, as described with reference to FIGS. 8A to 9, the clusters 3 (CLOS networks) can be automatically connected to each other.

This is the end of the description of the embodiments, but the aspects of the present invention are not limited to these.
For example, in the present embodiment, a connection using a physical interface is assumed as the simplest example, but in practice, a virtual interface such as a VLAN (Virtual Local Area Network) or LAG (Link Aggregation) is used to connect to the ground. Sometimes connected. In this case, if the above-described virtual interface is managed instead of the physical IF in the logical IF information table 91 (FIG. 6B (a)) and the logical IF information table 91a (FIG. 8B (a)), the same technique is used. Can be realized.
In addition, about a concrete structure, it can change suitably in the range which does not deviate from the main point of this invention.

1 Host controller (network management device)
2 Backbone Network 3 Cluster 5 Switch Model Information 6 Intra-cluster Device Information Table 7 WAN Information 8 Cluster Information 12 Storage Unit 21 Relay Device 31 Controller (CLOS Network Management Device)
32 Spine Switch 33 Leaf Switch 34 Server 35 Switch 91 Logical IF Information Table 92 Connection Information Table 312 Setting Control Unit 313 DHCP Server Unit 314 Syslog Server Unit

Claims (6)

In a network including a plurality of CLOS networks having a plurality of Spine devices and a plurality of Leaf devices, a network management device that performs address setting of the Spine device and the Leaf device,
A starting point address which is a starting point of an address set in the Spine device and the Leaf device, and a storage unit for storing the number of the Spine devices and the number of the Leaf devices for each CLOS network;
Based on the starting point address and the number of the Spine devices and the number of Leaf devices for each CLOS network, a starting point address for each CLOS network that is the starting point of the address for each CLOS network is determined,
For each CLOS network, a processing unit that sets the addresses of the Spine device and the Leaf device based on the start address for each CLOS network, and the number of the Spine devices and the number of Leaf devices;
Equipped with a,
The processor is
Based on the descending order of the product of the number of Spine devices and the number of Leaf devices for each CLOS network, the cluster ID of the CLOS network is determined,
The starting address for each CLOS network is determined so that the descending order is arranged in order from the CLOS network,
The addresses of the Spine device and Leaf device of the one CLOS network are set after the network-specific start address of the CLOS network and before the network-specific start address of the CLOS network of the next cluster ID. A network management device characterized by the above. A CLOS network management apparatus installed for each CLOS network managed by the network management apparatus according to claim 1,
If any of the Spine device and the Leaf device is added in its own CLOS network,
Based on the number of Spine devices and the number of Leaf devices in its own CLOS network, the address of the device to be added is determined,
Notify the determined address to the added device,
CLOS network management characterized in that among the existing Spine device and Leaf device, setting information for communication with the added device is notified to a device connected to the added device. apparatus. The storage unit stores a CLOS network start point interface address which is a start point for determining a CLOS network-specific interface address used for connection between the CLOS networks,
The processor is
Based on the inter-CLOS network starting point interface address, and the number of Spine devices and the number of Leaf devices for each CLOS network, determine the interface address for each CLOS network,
The determined interface address for each CLOS network is transmitted to each CLOS network management device installed for each CLOS network, and the CLOS network management device is notified to the Spine device and the Leaf device under its control, The network management apparatus according to claim 1, wherein the CLOS networks are connected to each other. In a network including a plurality of CLOS networks having a plurality of Spine devices and a plurality of Leaf devices, an address setting method by a network management device that performs address setting of the Spine device and the Leaf device,
The network management device includes:
A starting point address that is a starting point of an address set in the Spine device and the Leaf device, a storage unit that stores the number of the Spine devices and the number of Leaf devices for each CLOS network, and a processing unit. And
The processor is
Based on the descending order of the product of the number of Spine devices and the number of Leaf devices for each CLOS network, the cluster ID of the CLOS network is determined,
Based on the starting point address and the number of Spine devices and the number of Leaf devices for each CLOS network, the starting point address for each CLOS network that is the starting point of the address for each CLOS network is the CLOS network with the highest descending order. Decided to line up in order ,
For each CLOS network, based on the start address for each CLOS network, and the number of Spine devices and the number of Leaf devices, the addresses of the Spine device and Leaf device of one CLOS network are The address of the Spine device and the Leaf device is set so that it is after the network-specific start address and before the network-specific start address of the CLOS network of the next cluster ID. Method. An address setting method by a CLOS network management device installed for each CLOS network managed by the network management device performing the address setting method of claim 4,
The CLOS network management device
If any of the Spine device and the Leaf device is added in its own CLOS network,
Based on the number of Spine devices and the number of Leaf devices in its own CLOS network, the address of the device to be added is determined,
Notify the determined address to the added device,
Of the existing Spine device and the Leaf device, a device connected to the added device is notified of setting information for communication with the added device. . The storage unit stores a CLOS network start point interface address which is a start point for determining a CLOS network-specific interface address used for connection between the CLOS networks,
The processor is
Based on the inter-CLOS network starting point interface address, and the number of Spine devices and the number of Leaf devices for each CLOS network, determine the interface address for each CLOS network,
The determined interface address for each CLOS network is transmitted to each CLOS network management device installed for each CLOS network, and the CLOS network management device is notified to the Spine device and the Leaf device under its control, The address setting method according to claim 4, wherein the CLOS networks are connected to each other.

Images