A DPU-based cloud host live migration network scheme

Abstract

The application discloses a cloud host network migration scheme based on DPU, and comprises the following steps: S1, basic network configuration; S2, out direction traffic identification and forwarding flow table configuration; and S3, in direction traffic identification and forwarding configuration. The scheme does not need additional planning calculation of private IP addresses, does not involve SNAT, and does not additionally design vpc and overlay, so that network migration channels are realized by utilizing PF and reusing original calculation internal network addresses, and the bandwidth of the migration network is ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of emerging information technology, and in particular to a cloud host live migration network scheme based on DPU. BACKGROUND

[0002] In the cloud computing virtualization scenario, in order to pursue the ultimate physical server CPU and memory utilization, and release 20% of the computing power tax, a cloud host scheme based on DPU is adopted. The internal network topology of the non-DPU cloud host is that the management port, the computing internal network, and the storage network are all on the computing node system, that is, the cloud host business traffic, management traffic, network migration traffic, and storage traffic are all communicated with the outside through the network cards in different network planes in the computing node system.

[0003] The DPU cloud host network topology is that only the management port is on the computing node, the computing internal network and the storage network are on the DPU, and the cloud host business traffic and the storage traffic are all transmitted to the DPU through the VF, and then are transmitted out through the computing internal network and the storage network on the DPU. The VF communication mode is a virtual interface simulated by the front-end and back-end virtio-net, one end of which is exposed in the cloud host system, and the other end is exposed on the DPU, which is called a representative port. The PF of the DPU is not utilized in the cloud host scenario.

[0004] Under this background, the DPU cloud host migration scenario has limitations. In the case that the management port of the computing node is of small bandwidth, the cloud host migration is not suitable for migrating through the management network, because the management network also has other management traffic, such as vnc and computing node management. Therefore, the PF of the DPU needs to be utilized to serve as a PF representative port to migrate through the DPU virtual bridge.

[0005] The existing PF migration network scheme includes adding a new network channel, SNAT, and Overlay. The new network channel is to reserve the computing internal network IP address from the resource pool network planning stage. The migration traffic is distributed out through the PF representative port on the br-ext bridge of the DPU through the flow table. The SNAT scheme is to plan a private IP address for the PF. The PF representative port on the DPU performs Full NAT, that is, converts the source and destination private IP addresses into the source and destination addresses of the computing internal network. The Overlay mode is the same as the SNAT scheme, which also needs to reserve a private IP address. The PF representative port is added to the br-int bridge of the DPU through the ovs controller, and the flow table is used to pass through the vxlan tunnel. This scheme can be compared to creating a migration vpc for the resource pool. The private IP address allocated on each computing node is parallel to the IP of the created cloud host. The ovs controller manages the life cycle and forwarding flow table of the PF representative port as an elastic network card.

[0006] In summary, the disadvantages in the prior art are:

[0007] The newly added computing internal network IP address threatens the IPv4 address pool resources, and there is a risk of insufficient address for the expansion of the existing DPU cloud host, which also brings more workload to the address planning group, and increases the complexity of the network plane, which is contrary to the trend of simplifying the physical network;

[0008] The SNAT scheme causes port conflicts and has low performance;

[0009] The OVERLAY scheme needs an additional adaptation controller, and the design migrates vpc, which also wastes vpc and ip address resources;

[0010] The migration network shares bandwidth with the computing network or the management network, and the bandwidth cannot be guaranteed. SUMMARY

[0011] This section is intended to summarize some aspects of the embodiments of the present application and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract and title of the specification to avoid obscuring the purpose of this section, the abstract and the title, and such simplifications or omissions cannot be used to limit the scope of the present application.

[0012] In view of the above problems of the prior art, the present application is proposed.

[0013] Therefore, the purpose of the present application is to provide a DPU-based cloud host hot migration network scheme, which is suitable for solving the problem of insufficient bandwidth of the VM migration public management port under the DPU cloud host product.

[0014] To solve the above technical problems, the present application provides the following technical scheme: a DPU-based cloud host hot migration network scheme, comprising the following steps:

[0015] S1: basic network configuration;

[0016] S2: out direction traffic identification, forwarding flow table configuration;

[0017] S3: in direction traffic identification, forwarding configuration.

[0018] As a preferred scheme of the DPU-based cloud host hot migration network scheme, wherein: in step S1, it comprises the following sub-steps:

[0019] Step one: add PF representative port pf0hpf to br-ext bridge on DPU;

[0020] Step two: set the PF port state of the host side to up;

[0021] Step three: host side PF configure and DPU the same ip address and mac address as the compute network.

[0022] As a preferred solution of the DPU-based cloud host live migration network solution of the application, in step S2, the br-ext bridge root in_port = pf0hpf, tp_dst = <migrate_port> identifies the migration traffic.

[0023] As a preferred solution of the DPU-based cloud host live migration network solution of the application, in step S2, the br-ext bridge root in_port = pf0hpf, tp_dst = <migrate_port> identifies the migration traffic.

[0024] As a preferred solution of the DPU-based cloud host live migration network solution of the application, in step S3, the br-ext bridge root in_port = bond1, dl_vla = 301, tp_src = <migrate_port> identifies the migration traffic.

[0025] As a preferred solution of the DPU-based cloud host live migration network solution of the application, in step S3, the br-ext bridge root in_port = bond1, dl_vla = 301, tp_src = <migrate_port> identifies the migration traffic.

[0026] As a preferred solution of the DPU-based cloud host live migration network solution of the application, the network solution further comprises S4: enabling dscp-based pfc.

[0027] # mlnx_qos -i p0

[0028] DCBX mode: OS controlled

[0029] Priority trust state: dscp

[0030] dscp2prio mapping:

[0031] prio:0 dscp:07,06,05,04,03,02,01,00,

[0032] prio:1 dscp:15,14,13,12,11,10,09,08,

[0033] prio:2 dscp:23,22,21,20,19,18,17,16,

[0034] prio:3 dscp:31,30,29,28,27,26,25,24,

[0035] prio:4 dscp:39,38,37,36,35,34,33,32,

[0036] prio:5 dscp:47,46,45,44,43,42,41,40,

[0037] prio:6 dscp:55,54,53,52,51,50,49,48,

[0038] prio:7 dscp:63,62,61,60,59,58,57,56,

[0039] default priority:

[0040] Receive buffer size(bytes):20016,156096,0,0,0,0,0,0,

[0041] Cable len:7

[0042] PFC configuration:

[0043]

[0044] tc:0 ratelimit:unlimited,tsa:vendor

[0045] priority:1

[0046] tc:1 ratelimit:16.0 Gbps,tsa:vendor

[0047] priority:0

[0048] tc:2 ratelimit:unlimited,tsa:vendor

[0049] priority:2

[0050] tc:3 ratelimit:8.0 Gbps,tsa:vendor

[0051] priority:3

[0052] tc:4 ratelimit:2.0 Gbps,tsa:vendor

[0053] priority: 4

[0054] tc: 5 ratelimit: unlimited, tsa: vendor

[0055] priority: 5

[0056] tc: 6 ratelimit: unlimited, tsa: vendor

[0057] priority: 6

[0058] tc: 7 ratelimit: unlimited, tsa: vendor

[0059] priority: 7.

[0060] As a preferred scheme of the DPU-based cloud host live migration network scheme, wherein: the DPU adopts two 25G ten-gigabit port specifications, and two PFs on the DPU are bonded 4 on the system.

[0061] As a preferred scheme of the DPU-based cloud host live migration network scheme, wherein: by marking the traffic with a high-priority dscp value, there is a corresponding high-priority queue on the physical port of the DPU, a specific bandwidth is reserved, and the migration traffic cannot be preempted on the export.

[0062] As a preferred scheme of the DPU-based cloud host live migration network scheme, wherein: in step S3, in the incoming direction, the vxlan tunnel service traffic is distinguished by in_port=bond1, tp_dst=4789, priority=high, and the migration network is distinguished by in_port=bond1, tp_src=migrate_port, priority=middle.

[0063] The application has the beneficial effects that: the PF port on the host side is used for migration network, the limitation of small management network bandwidth is broken through, the physical network is simplified, and no additional network address planning and network plane are involved;

[0064] No nat and overlay mechanism are involved, the unerlay performance is high, the computing internal network and the computing internal network IP address are multiplexed, the pfc is based on dscp, and the bandwidth is guaranteed. BRIEF DESCRIPTION OF DRAWINGS

[0065] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description only represent some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings. Among them:

[0066] Fig. 1 A DPU representative diagram of a cloud host live migration network scheme based on the DPU is provided for the present application.

[0067] Fig. 2 A whole framework structure diagram of the cloud host live migration network scheme based on the DPU is provided for the present application.

[0068] Fig. 3 A flow step diagram of the cloud host live migration network scheme based on the DPU is provided for the present application. DETAILED DESCRIPTION

[0069] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings.

[0070] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0071] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments.

[0072] Thirdly, the present application is described in detail in conjunction with the schematic diagram, and in the detailed description of the embodiments of the present application, the cross-sectional view of the device structure is partially enlarged without general proportion for the convenience of description, and the schematic diagram is only an example, which should not limit the scope of protection of the present application herein. In addition, three-dimensional spatial dimensions including length, width and depth should be included in actual manufacture.

[0073] EMBODIMENT

[0074] REFERENCE Figs. 1-3 For one embodiment of the present application, a cloud host live migration network scheme based on DPU is provided, including the following steps:

[0075] S1: basic network configuration;

[0076] In step S1, it includes the following sub-steps:

[0077] Step one: add PF representative port pf0hpf to the br-ext bridge on the DPU, the DPU adopts two 25G specifications, two PFs on the DPU are aggregated as BOND4 on the system, by marking high priority dscp value to the traffic, there is a corresponding high priority queue on the physical port of the DPU, reserving specific bandwidth, guaranteeing that the migration traffic will not be preempted on the export;

[0078] Step two: set the PF port state on the host side as up;

[0079] Step three: the host side PF configures the same ip address and mac address as the computing network of the DPU;

[0080] S2: out direction traffic identification, forwarding flow table configuration;

[0081] The br-ext bridge identifies the migration traffic according to in_port=pf0hpf, tp_dst=<migrate_port>, and then sends it out from bond1 by marking the computing intranet vlan;

[0082] S3: in direction traffic identification, forwarding configuration;

[0083] The br-ext bridge identifies the migration traffic according to in_port=bond1, dl_vla=301, tp_src=<migrate_port>, and then sends it out from the PF representative port by stripping the vlan;

[0084] In the in direction, the vxlan tunnel service traffic is distinguished by in_port=bond1, tp_dst=4789, priority=high, and the migration network is distinguished by in_port=bond1, tp_src=migrate_port, priority=middle.

[0085] The network scheme further includes S4: enabling dscp-based pfc:

[0086] #mlnx_qos-i p0

[0087] DCBX mode:OS controlled

[0088] Priority trust state:dscp

[0089] dscp2prio mapping:

[0090] prio:0 dscp:07,06,05,04,03,02,01,00,

[0091] prio:1 dscp:15,14,13,12,11,10,09,08,

[0092] prio:2 dscp:23,22,21,20,19,18,17,16,

[0093] prio:3 dscp:31,30,29,28,27,26,25,24,

[0094] prio:4 dscp:39,38,37,36,35,34,33,32,

[0095] prio:5 dscp:47,46,45,44,43,42,41,40,

[0096] prio:6 dscp:55,54,53,52,51,50,49,48,

[0097] prio:7 dscp:63,62,61,60,59,58,57,56,

[0098] default priority:

[0099] Receive buffer size(bytes):20016,156096,0,0,0,0,0,0,

[0100] Cable len:7

[0101] PFC configuration:

[0102]

[0103] tc:0 ratelimit:unlimited,tsa:vendor

[0104] priority:1

[0105] tc:1 ratelimit:16.0 Gbps,tsa:vendor

[0106] priority:0

[0107] tc:2 ratelimit:unlimited,tsa:vendor

[0108] priority:2

[0109] tc:3 ratelimit:8.0 Gbps,tsa:vendor

[0110] priority:3

[0111] tc:4 ratelimit:2.0 Gbps,tsa:vendor

[0112] priority:4

[0113] tc:5 ratelimit:unlimited,tsa:vendor

[0114] priority:5

[0115] tc:6 ratelimit:unlimited,tsa:vendor

[0116] priority:6

[0117] tc:7 ratelimit:unlimited,tsa:vendor

[0118] priority:7.

[0119] In use, by adding the representative port of the PF to the br-ext bridge of the DPU, the host side main network card is configured with the same calculation IP and mac address as on the DPU, different flows are distinguished on the br-ext bridge by configuring a flow table, the migration flow is identified by in_port=pf and tp_port=migrate_port in the out direction, the migration flow is marked with a high priority dscp value, there is a corresponding high priority queue on the physical port of the DPU, a specific bandwidth is reserved, the migration flow is guaranteed not to be preempted on the export, the DPU and the host side public calculation intranet are used to realize cloud host migration, the dscp-based pfc flow control technology is used to guarantee the bandwidth of the migration network.

[0120] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present application, and they should be covered in the scope of the claims of the present application.

Claims

1. A DPU-based cloud host live migration network method, characterized in that, Comprising the following steps: S1: base network configuration; S2: out direction traffic identification, forwarding table configuration; S3: in direction traffic identification, forwarding configuration; In step S1, it comprises the following sub-steps: Step one: add PF representative port pf0hpf to br-ext bridge on DPU; Step two: set PF port status as up on host side; Step three: host side PF configuration and DPU's compute network same IP address and mac address; In step S2, br-ext bridge root in_port=pf0hpf, tp_est=<migrate_port> identifies migration traffic; In step S2, by giving it compute intranet's vlan, then send it out from bond1; In step S3, br-ext bridge root in_port=bond1, dl_vla=301, tp_src=<migrate_port> identifies migration traffic; In step S3, by stripping its vlan, send it out from PF's representative port. The network method further comprises S4: enabling dscp-based pfc: #mlnx_qos -i p0 2.The DPU-based cloud host live migration network method of claim 1, wherein: DCBX mode: OS controlled Priority trust state: dscp dscp2prio mapping: prio:0 dscp:07,06,05,04,03,02,01,00, prio:1 dscp:15,14,13,12,11,10,09,08, prio:2 dscp:23,22,21,20,19,18,17,16, prio:3 dscp:31,30,29,28,27,26,25,24, prio:4 dscp:39,38,37,36,35,34,33,32, prio:5 dscp:47,46,45,44,43,42,41,40, prio:6 dscp:55,54,53,52,51,50,49,48, prio:7 dscp:63,62,61,60,59,58,57,56, default priority: Receive buffer size (bytes): 20016,156096,0,0,0,0,0,0, Cable len: 7 PFC configuration: priority 0 1 2 3 4 5 6 7 enabled 0 0 0 1 0 0 0 0 buffer 0 0 0 1 0 0 0 0 tc: 0 ratelimit: unlimited, tsa: vendor priority: 1 ​ ​ tc: 1 ratelimit: 16.0 Gbps, tsa: vendor priority: 0 tc: 2 ratelimit: unlimited, tsa: vendor priority: 2 tc: 3 ratelimit: 8.0 Gbps, tsa: vendor priority: 3 tc: 4 ratelimit: 2.0 Gbps, tsa: vendor priority: 4 tc: 5 ratelimit: unlimited, tsa: vendor priority: 5 tc: 6 ratelimit: unlimited, tsa: vendor priority: 6 tc: 7 ratelimit: unlimited, tsa: vendor priority:

7. 3.The DPU-based cloud host live migration network method of claim 1, wherein: The DPU adopts two 25G specifications, and two PFs on the DPU are bonded on the system.

4. The DPU-based cloud host live migration network method of claim 1, wherein: By marking the traffic with a high-priority dscp value, there will be a corresponding high-priority queue on the physical port of the DPU, reserving a specific bandwidth to ensure that the migration traffic will not be preempted on the export.

5. The DPU-based cloud host live migration network method of claim 1, wherein: In step S3, in the incoming direction, the vxlan tunnel service traffic is distinguished by in_port=bond1, tp_dst=4789, priority=high, and the migration network is distinguished by in_port=bond1, tp_src=migrate_port, priority=middle.

Images