Infrastructure level LAN security

Abstract

Techniques are disclosed for securing traffic flowing across multi-tenant virtualized infrastructures using group key-based encryption. In one embodiment, an encryption module of a virtual machine (VM) host intercepts layer 2 (L2) frames sent via a virtual NIC (vNIC). The encryption module determines whether the vNIC is connected to a “secure wire,” and invokes an API exposed by a key management module to encrypt the frames using a group key associated with the secure wire, if any. Encryption may be performed for all frames from the vNIC, or according to a policy. In one embodiment, the encryption module may be located at a layer farthest from the vNIC, and encryption may be transparent to both the VM and a virtual switch. Unauthorized network entities which lack the group key cannot decipher the data of encrypted frames, even if they gain access to such frames.

Description

BACKGROUND

Cloud computing is a service model where computing resources are delivered over a network. Typically, a common virtualized infrastructure is used to provide services for multiple client organization (tenants). Advantages of cloud computing include lower costs through shared computing and network infrastructure, on-demand provisioning of resources, and multi-tenant solutions, among others.

Data security within the shared cloud network infrastructure is a major concern. Although secure web services and virtual private networks (VPNs) can provide secure communication with the cloud, such security ends at the entry point to the cloud provider. That is, data traversing the internal network of the cloud provider's infrastructure typically flows in an unsecure manner.

One solution for securing data traversing the provider's internal network uses virtual local area networks (VLANs), which afford tenants segregated logical networks. However, VLANs may allow malicious entities to gain unauthorized access to network data by means such as passive wiretapping, masquerading attacks, man-in-the-middle attacks, private VLAN attacks, and some denial of service attacks.

Another solution for securing data traversing the provider's internal network relies on point-to-point encryption techniques, such as Secure Sockets Layer (SSL) or IPSec. However, such techniques, which typically require negotiation of encryption keys by each pair of endpoints wishing to communicate securely, do not scale well. For example, if n virtual machines (VMs) (or applications running thereon) wish to communicate securely with each other, then each VM (or application) must be capable of performing the particular type of encryption (e.g., SSL encryption), as well as negotiate and manage n*(n−1) security keys.

SUMMARY

Embodiments presented herein provide techniques for securing traffic in a multi-tenant virtualized infrastructure. The techniques include intercepting a Layer 2 (L2) frame sent via a first virtual network interface card (vNIC) in route to a first physical network interface card (pNIC) and determining a first secure wire to which the first vNIC is connected, the first secure wire being enabled on a first L2 domain. The techniques further include encrypting payload data of the L2 frame using a first encryption key associated with the first secure wire.

Further embodiments of the present invention include a computer-readable storage medium storing instructions that when executed by a computer system cause the computer system to perform one or more the of the techniques set forth above, and a computer system programmed to carry out one or more of the techniques set forth above.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 illustrates a system in which embodiments may be implemented.

FIG. 2 illustrates an example encrypted Ethernet frame, according to an embodiment of the invention.

FIG. 3 illustrates another example encrypted Ethernet frame, according to an embodiment of the invention.

FIG. 4 illustrates an encrypted virtual extensible LAN (VXLAN) packet, according to an embodiment of the invention.

FIG. 5 illustrates a method for encrypting and transmitting data over a secure wire, according to an embodiment of the invention.

FIG. 6 illustrates a method for creating a secure wire and adding network interface cards (vNICs) to the secure wire, according to an embodiment of the invention.

DETAILED DESCRIPTION

Embodiments presented herein secure traffic flowing across multi-tenant virtualized infrastructures using group key-based encryption while being transparent to virtual machines (VMs) and application(s) running within VMs. In one embodiment, an encryption module in a VM host intercepts Layer 2 (L2) frames i.e., Ethernet frames which a VM sends via its virtual network interface card (vNIC). The encryption module then determines whether the vNIC is connected to a “secure wire.” As used herein, “secure wire” is an infrastructure-level construct over which at least some L2 traffic is encrypted after leaving source vNIC(s) and decrypted before reaching target vNIC(s). A group encryption key may be associated with the secure wire such that entities (e.g., VMs and physical servers) belonging to one secure wire context share the same encryption key. Upon determining that a vNIC, from which L2 frames are sent, is connected to a given secure wire, the encryption module invokes an API exposed by a key management module to encrypt the frame data. Encryption keys are kept local to the key management module, which is responsible for fetching the keys from a centralized key manager and which exposes the API for encrypting frame data using the keys. Alternatively, key management module may use hardware chips to perform cryptographics operations in a manner that is transparent to encryption module. Frames having encrypted payloads and additional fields are also referred to herein as “encrypted frames.” In one embodiment, all traffic flowing over the secure wire may be encrypted. In alternative embodiments, the policy may specify which traffic to encrypt and the encryption module may intercept only the specified traffic for encryption. For example, the policy may specify encrypting only packets flowing from a VM with IP address X and destined to an HTTP server on IP address Y and port 80, securing traffic from address X to address Y but permitting address Y to continue to receive unencrypted frames from other VMs. In a further embodiment, the encryption module may be located at a layer farthest from a vNIC of the VM. That is, other VM host modules which act on the L2 frames (e.g., by applying policy decisions or application level modifications) are at layers closer to the vNIC and can thus process the L2 frames before those frames are passed on to the encryption module. This ensures that the functionality of the other VM host modules are unaffected by the L2 frame encryption. Embodiments may also support IEEE 802.1AE MAC Security (MACsec) standard, Group Domain of Interpretation (GDOI), and the like so that secure wires may stretch to include physical devices.

The encryption module forwards encrypted frames to a physical network interface card, which transmits the encrypted frames via a physical layer network medium. Unauthorized network entities (i.e., entities not connected to the secure wire) that gain access to encrypted frames cannot decipher the frames' payloads. A decryption module in a VM host supporting a VM which includes the target vNIC may intercept the encrypted frame before the frame reaches the target vNIC. The decryption module may then decrypt the encrypted data of the frame and forward the decrypted frame to the target vNIC. In one embodiment, the decryption module may be located at a layer farthest from a vNIC of the receiving VM, such that other VM host modules may act on the decrypted L2 frames before the frames are passed to the target vNIC.

Reference is now made in detail to several embodiments, examples of which are illustrated in the accompanying figures. It is noted that wherever practicable similar or like reference numbers may be used in the figures and may indicate similar or like functionality. The figures depict embodiments for purposes of illustration only. One skilled in the art will readily recognize from the following description that alternative embodiments of the structures and methods illustrated herein may be employed without departing from the principles described herein.

FIG. 1 illustrates a system in which embodiments may be implemented. As shown, the system includes host servers 101_1-2, key management server 150, and virtual extensible LAN (VXLAN) backed virtual wire 160. Host servers 101_1-2 provide virtualization-based distributed services to information technology environments. Illustratively, host servers 101_1-2 each support a virtual machine host 120_1-2. Each virtual machine host 120_1-2 may itself include virtualization software that shares hardware resources (e.g., processors, memory, disks, etc.) among virtual machines and provides an interface between guest software running on virtual machines 110₁and 110₂, and the underlying hardware of host servers 101₁ and 101₂. VM hosts 120_1-2 may run directly on hardware components of host servers 101_1-2 or on top of an operating system of host servers 101_1-2. In a particular embodiment, VM hosts 120_1-2 may be VMware ESX® Servers. Note, host servers 101₁ and 101₂ may belong to different physical subnetworks. As a result, packets sent between host servers 101_1-2 may flow over intermediate network devices that could otherwise inspect and/or alter the packets. Techniques disclosed herein encrypt packet data, permitting VMs 110_1-2 to communicate securely.

As shown, VM hosts 120_1-2 each include a virtual network interface card (vNIC) emulator 121_1-2 which interacts with NIC drivers 116_1-2 in VMs 110_1-2, to send and receive data to and from VMs 110_1-2. In particular, each vNIC emulator 121_1-2 may maintain the state for one or more vNICs for each VM 110_1-2. Alternatively, multiple instances of vNIC emulators may be instantiated within a virtualization software layer of VM hosts 120_1-2. In either case, a single VM may have one or more vNICs. Illustratively, FIG. 1 shows one vNIC for each VM, and only one VM for each host. However, the discussion of vNICs 118_1-2 actually describes a vNIC state implemented and maintained by each vNIC emulator 121_1-2, respectively. Virtual devices such as vNICs 118_1-2 are software abstractions convenient to discuss as though part of VMs 110_1-2, but are actually implemented by virtualization software of VM hosts 120_1-2 using emulators 121_1-2. The state of each VM 110_1-2, however, includes the state of its virtual devices, which is controlled and maintained by the underlying VM hosts 120_1-2.

As also shown, the VM hosts 120_1-2 each include a component of a distributed virtual switch (DVSwitch) 130. The DVSwitch 130 is a software abstraction which binds virtual switches in a managed collection into a single, logical configurable entity. Each of the ports of the DVSwitch 130, referred to herein as DVPorts, may include data structures representing the configuration and runtime states of a virtual port of a virtual switch supported by a VM host. DVPorts may be created with, e.g., a configuration predefined by a network administrator. Virtual ports are initialized with a blank configuration state, but once associated with a DVPort, assume the configuration and runtime state of the associated DVPort. When a VM is migrated or powered off and on, the “connection” between a DVPort and a virtual NIC is not affected because the DVPort persists and migrates with the VM to which it is connected. A further description of DVSwitches is provided in U.S. Pat. No. 8,195,774 B2, which is hereby incorporated by reference in its entirety. Note, in alternative embodiments, switches other than DVSwitches (i.e., other types of virtual switches) may be used.

Encryption module 125 intercepts L2 frames sent via vNIC 118₁, invokes an API 127₁ exposed by a key management module 126₁, discussed in greater detail below, to encrypt the frames' data, then forwards the encrypted frames to a physical NIC 143₁. As shown, encryption module 125 may be located at a layer between DVSwitch 130 and physical network interface card (pNIC) 143₁, such that secure wire is transparent to DVSwitch 130 and VM 110₁. That is, DVSwitch 130 and VM 110₁ may be unaware of the secure wire and the encrypted frames transmitted over the secure wire. The intercepted L2 frames such as Ethernet frame 115 each encapsulate an Internet Protocol (IP) packet which itself encapsulates a Hypertext Transfer Protocol (HTTP) request (or other application layer packet). After intercepting the L2 frames from vNIC 118₁, encryption module 125 may determine which secure wire, if any, vNIC 118₁ belongs to, and invoke the API 127₁ to encrypt the frames. That is, secure wire membership may be derived based on the membership of vNIC 118₁, from which the packet has arrived, such that if vNIC 118₁ belongs to a given secure wire, the encryption module 125 takes further action(s) to encrypt the packets. In an alternative embodiment, secure wire membership may be derived based on policies. That is, the encryption module 125 may inspect packets and decide secure wire membership based on packet characteristics, as discussed in greater detail below.

Each secure wire is associated with a group encryption key. Group encryption keys may be managed via a key manager application 151 of a management server 150. Application(s) running on the management server 150 may generate the encryption key when the secure wire is created. For example, a user may create a virtual extensible LAN (VXLAN) backed virtual wire 160 via an interface provided by a network management platform 149 and set secure wire as “enabled” on virtual wire 160. The network management platform 149 may then program the key manager application 151 on the management server to generate the encryption key. In a particular embodiment, the network management platform 149 may be a VMware vShield™ Manager, and the key manager application 151 may be Key Management Interoperability Protocol (KMIP) compliant. In an alternative embodiment, the user may also configure manual group keys on the virtualization management platform if no management server 150 is available.

In general, VXLAN permits L2 logical networks to be created with a 24-bit segment ID used to differentiate the L2 networks. In addition, VXLAN uses MAC in IP encapsulation to stretch L2 networks across layer 3 (L3) boundaries. In particular, Ethernet frames from VMs may be encapsulated inside UDP, with the UDP header used to traverse L3 boundaries. A user may wish to enable secure wire on VXLAN backed virtual wire 160 so that Ethernet frame data is encrypted before the frames are encapsulated in UDP. Illustratively, encryption module 125 may invoke API 127₁ to encrypt frame data, and a further virtualization software module (not shown) may create VXLAN packets (e.g., packet 155) which encapsulate the encrypted frames. The VXLAN packets are then forwarded to a pNIC 143₁ for transmission over VXLAN backed virtual wire 160.

As discussed, a key manager application 151 may manage group encryption keys. Key management modules 126_1-2 in virtualization software 120_1-2 may securely fetch the appropriate keys from the key manager application 151 when vNICs 118_1-2 join secure wire(s), respectively. For example, the key manager application 151 may use SSL or similar point-to-point security techniques to push keys to key management modules 126_1-2. In one embodiment, key management modules 126_1-2 may manage keys for virtualization software 120_1-2, respectively, and expose a framework of cryptographic libraries which encryption module 125 and decryption module 126 consume. For example, encryption module 125 may create a session per secure wire, and key management module 126₁ may fetch the appropriate key for a newly created secure wire and associate the key with the secure wire. Here, group keys may expire after a given time, and in such cases the keys may be refreshed periodically. In a particular embodiment, the key management module 126₁ may be a VMware vCrypto® module. In another embodiment, key manager application 151 may be a key management interoperability protocol (KMIP) server application, and the key management module 126₁ may negotiate an SSL channel with the KMIP server to retrieve group keys. In yet another embodiment, the key manager application 151 may be a Group Domain of Interpretation (GDOI) server application, and key management module 126₁ may communicate with key manager application 151 via a secure channel negotiated over Internet Key Exchange (IKE) phase 1.

Assuming encryption keys are available via the key management module 126₁, the encryption module 125 invokes an API 127₁ exposed by key management module 126₁to encrypt Ethernet frame data from vNIC 118₁. Here, the API 127₁ may accept plain text data and return cipher text. As a result, encryption keys are kept local to key management module 126₁ and not available outside its boundary. That is, the key management module 126₁ performs group key negotiation and actual encryption of frames. On the other hand, the encryption module 125 is responsible for intercepting traffic, and uses interfaces provided by the key management module 126₁ to encrypt the traffic. In an alternative embodiment, the key management module 126₁ may use hardware chips to perform cryptographics operations in a manner that is transparent to encryption module 125.

Of course, encryption keys may be managed and used in other ways. For example, encryption module 125 may itself manage keys and perform encryption using the keys. In any case, encrypted frames may be encapsulated in a VXLAN packet and sent over virtual wire 160. In one embodiment, data of Ethernet frames sent via vNIC 118₁ may be encrypted using an appropriate key if vNIC 118₁ is connected to a secure wire. In alternative embodiments, encryption may be policy-based. That is, Ethernet frames may be selectively encrypted based on policies specifying which traffic to encrypt. For example, the policy may specify that frames with a given source and/or destination IP address are encrypted. As another example, the policy may specify that encryption is performed at higher layers (e.g., OSI Layer 3 (L3) or Layer 4 (L4)), as discussed in greater detail below.

As shown, the encryption module 125 is located at a layer farthest from the vNIC 118₁. This configuration permits other modules 122₁ which apply logic (e.g., policy decisions or application level modifications) to the L2 frames to see clear traffic. Note, in some embodiments (e.g., MACSEC encryption), the encryption module 125 may be above the VXLAN layer, while in other embodiments (e.g., GDOI) the encryption module 125 may be below the VXLAN layer.

On reaching VM host 120₂, the VXLAN packet which includes the encrypted packet is stripped of its VXLAN header and forwarded to vNIC 118₂ based on the destination MAC address specified in the frame. Before reaching vNIC 118₂, a decryption module 126 intercepts the frame and invokes API 127₂, exposed by key management module 126₂, to decipher the frame data using the key indicated in a Key ID field of the frame. Similar to the key management module 126₁, the key management module 126₂ may perform group key negotiation and actual decryption of frames. Decryption module 126 may then be responsible for intercepting traffic and using interfaces provided by the key management module 126₂ to decrypt the traffic. Note, in one embodiment, the decryption module 126 may also encrypt outgoing frames, and the encryption module 125 may also decrypt inbound frames.

Once decrypted, the decryption module 126 may forward frames to DVSwitch 130 for transmission to vNIC 118₂ of VM 110₂. Here, decryption may not actually be successful if vNICs 118₁ and 118₂ do not belong to the same secure wire, and thus do not share the same encryption key. In such a case, the key management module 126₂ may lack the key to decrypt the frame, and VM 110₂ will receive a frame which includes garbled data. Illustratively, the decryption module 126 is located at a layer farthest from vNIC 118₁. Similar to the placement of encryption module 125 at a layer farthest from vNIC 118₂, such a configuration permits other modules 122₂ which apply logic to the received L2 frames to see clear traffic.

Although discussed above relative to VXLAN, a secure wire may be used to provide security on other L2 domains. For example, a secure wire may be enabled for a distributed virtual port group (dvPG) associated with DVSwitch 130. In one embodiment, the scope of the secure wire may be inherited by underlying constructs of the L2 domain such that, e.g., when a secure wire is enabled on a VXLAN backed virtual wire, the secure wire is inherited by dvPGs of the virtual wire. As another example, when secure wire is enabled on a dvPG, the secure wire may be inherited by port groups (PGs) of the dvPG.

FIG. 2 illustrates an example encrypted Ethernet frame, according to an embodiment. As shown, a normal Ethernet frame 200 received from a vNIC includes MAC address fields 210, 220 and user data 230. The Ethernet frame 200 may also include, e.g., a preamble, a frame check sequence, and the like. In contrast, the encrypted Ethernet frame 250 includes the same MAC addresses 210, 220, as well as encrypted user data 285 and the following additional fields: EtherType 270, Key ID 280, Encoded Initialization Vector (IV) 290, and Signed Hash 295. Here, EtherType 270 may specify the type of protocol encapsulated in the payload as, e.g., “SecureWirePacket” or a similar designation. Key ID 280 and Encoded IV 290 are parameters used to encrypt the user data 230 to generate encrypted data 285. As discussed, the encryption module may invoke an API exposed by a key management module to encrypt frame data using a group encryption key which is unique to the particular secure wire. Key ID 280 may identify the encryption key used so that a decryption module on a receiving VM host may invoke an API to perform decryption based on the key identified by Key ID 280.

The actual encryption may be performed using the encryption key identified by the Key ID 280 and an initialization vector, such as a timestamp, which as illustrated is included in the Encoded IV 290 field of the encrypted frame. Advantageously, sending the initialization vector with the encrypted frame may avoid replay attacks, dictionary attacks, and the like. In one embodiment, the encryption module may encrypt data 230 using Galois/Counter mode of Advanced Encryption Standard (AES) cipher, with 128-bit encryption. Other encryption techniques may be applied and may also be indicated by Key ID 280.

Illustratively, the encrypted Ethernet frame 250 further includes a Signed Hash 295 field. The Signed Hash 295 may be a message digest (i.e., hash) encrypted using a key which is distinct from the data encryption key and generated at the creation of the secure wire. Similar to a digital signature, the Signed Hash 295 may authenticate the sender of the encrypted frame 250 and protect its data integrity by allowing changes to be detected.

FIG. 3 illustrates another example encrypted Ethernet frame, according to an embodiment. As shown, normal Ethernet frame 300 from the vNIC is similar to Ethernet frame 200, discussed above. However, the frame data is encrypted using IEEE 802.1AE Media Access Control Security (MACSec) format, which defines connectionless data confidentiality and integrity for media access independent protocols. Illustratively, the MACSec frame includes MAC addresses 355, 360, a Security Tag (SecTag) 370, secure data 380, and an Integrity Check Value (ICV) 390 authentication code. SecTag 370 is an extension of EtherType and includes a field indicating that MACSec frame 350 is of EtherType “MACSec.” Similar to Signed Hash 295, discussed above, ICV 390 is a message authentication code that permits authentication of the MACSec frame and protects data integrity of the frame. Similar to encrypted data 285, secure data 380 is user data 330 in encrypted format. MACSec provides a default cipher suite for encryption of user data and protection of data integrity, but MACSec does not specify how keys are managed and secure connections established. Thus, keys may be created and managed on a secure management server and pushed to key management modules of VM hosts, as discussed above.

Advantageously, MACSec is a standard protocol used for LAN encryption, and use of MACSec frame format permits physical end points which are MACSec compliant to participate in the secure wire. In the absence of MACSec compliant physical devices, an edge router (e.g., VMware's vShield Edge®) may facilitate participation of such physical devices in the secure wire.

FIG. 4 illustrates an encrypted VXLAN packet, according to an embodiment. As shown, normal VXLAN packet 400 includes an IP header 410 and IP payload 420. As discussed, VXLAN uses MAC-in-IP encapsulation. Here, IP payload 420 includes a UDP header 422, a VXLAN ID field 424, and a normal Ethernet frame 426 received from a vNIC. Illustratively, encrypted VXLAN packet 450 uses IPSec in GDOI format, which is configured in Encapsulating Security Payload (ESP) tunnel mode. As shown, IP header 410 is preserved in encrypted VXLAN packet 450, and encrypted VXLAN packet 450 further includes encrypted IP header 470 and encrypted IP payload 480. IP header 410 allows encrypted traffic to be routed using underlying network infrastructure. VXLAN also needs to support multicast traffic. Header preservation and group Security Association (SA) permit replication of multicast encrypted traffic to be offloaded to the underlying network. As discussed, GDOI may provide centralized encryption key management. In particular, GDOI key manager application may authenticate group members over IKE phase 1, distribute group SA to group members, distribute group policy specifying traffic to be encrypted, periodically refresh group keys of group members, and maintain a pseudo clock to protect against replay attacks.

FIG. 5 illustrates a method 500 for encrypting and transmitting data over a secure wire, according to an embodiment. As shown, the method 500 begins at step 510, where a first VM (VM 1) sends a frame via a first vNIC (vNIC 1). Here, vNIC 1 may be implemented by virtualization software of a VM host which supports VM 1, as illustrated in FIG. 1.

At step 520, the encryption module intercepts the frame as the frame is en route to a pNIC. As discussed, the encryption module may be a VM host module which is located at a layer farthest from the vNIC 1, thereby permitting other modules which apply logic to frames from vNIC 1 to see clear traffic.

At step 530, the encryption module determines which secure wire, if any, vNIC 1 is connected to. As discussed, secure wire membership of the vNIC may be maintained by the encryption module such that the encryption may intercept traffic and encrypt packets from vNICs which belong to secure wires, as appropriate. Initially, secure wire may be enabled on an L2 domain, and a user may thereafter assign vNICs to that domain. For example, the secure wire may be enabled on a management server, and a management server application may automatically generate encryption keys for new secure wires. A key management module on the VM host may securely fetch (e.g., using SSL or similar point-to-point security techniques) appropriate encryption keys when vNICs on the VM host join secure wires. The key management module may further maintain mappings between encryption keys and secure wires.

If the vNIC is connected to a secure wire, then at step 540, the encryption module invokes an API exposed by the key management module to encrypt the frames' payload using the key associated with the secure wire. As discussed, the API may accept plain text data and return cipher text in one embodiment. In such a case, encryption keys are kept local to key management module 126₁ and not available outside its boundary. The key management module 126₁ may apply any feasible encryption technique. For example, Galois/Counter mode of Advanced Encryption Standard (AES) cipher may be used. In an alternative embodiment, key management module 126₁ may use hardware chips to perform cryptographics operations in a manner that is transparent to encryption module 125. For example, some pNIC hardware chips support offloading of IPsec and MACsec encryption.

Additional fields may be added to the Ethernet frame such as a Key ID field identifying the encryption key and type of encryption, a timestamp initialization vector for the encryption, and a digital signature which authenticates the frame and protects data integrity of the frame. In one embodiment, the frame format of encrypted frames may be that illustrated in FIG. 2. In another embodiment, the encrypted frames may comply with the MACSec format, as illustrated in FIG. 3. In yet another embodiment, entire VXLAN packets may be encrypted, as illustrated in FIG. 4.

Although discussed above primarily with respect to all Ethernet frame payloads being encrypted if a vNIC is connected to a secure wire, the encryption module may apply encryption based on a policy in alternative embodiments. Tenants may choose policies which specify whether to perform encryption, particular traffic to encrypt, the secure wire(s) to which particular vNIC(s) belong, and the like in order to, e.g., improve performance or protect specific communications. Example policy rules are shown in Table 1.

TABLE 1
	Source	Destination	Source	Source	Destination	Destination
Index	Mac	MAC	IP	Port	IP	Port	Protocol	Action
1	ANY	ANY	X	ANY	Y	80	TCP	Encrypt
								to sec-
								wire-1
2	ANY	ANY	Y	80	X	ANY	TCP	Encryp
								to sec-
								wire-2
3	ANY	ANY	Z	80	ANY	ANY	TCP	Encrypt
								to sec-
								wire-1
4	ANY	ANY	ANY	ANY	Z	80	TCP	Encryp
								to sec-
								wire-3
5	ANY	ANY	ANY	ANY	ANY	443	ANY	Clear
6	ANY	ANY	ANY	443	ANY	ANY	ANY	Clear
7	ANY	ANY	ANY	ANY	ANY	ANY	ANY	Clear
8	ANY	FF:FF:FF:FF	ANY	ANY	ANY	ANY	ANY	Encrypt
								to sec-
								wire-1
9	ANY	ANY	ANY	ANY	ANY	ANY	ANY	Encrypt
								to sec-
								wire-3

As per rule 1, frames flowing from a VM having IP address X and destined to an HTTP server on IP address Y and port 80 are encrypted to secure wire 1, thereby securing HTTP traffic flowing from address X to address Y. Similarly, rule 2 secures return frames flowing from Y to X by encrypting the frames to secure wire 2. Note, given rules 1 and 2, Y may continue to honor unencrypted HTTP transfers with other VMs. If HTTP server on IP address Z should be allowed to communicate only securely with all VMs, then rules 3 and 4 may be applied instead. Rules 5 and 6 permit traffic to flow without encryption for HTTPS servers. Such servers already communicate securely, so no further encryption is necessary. Adding rule 7 permits broadcast traffic to flow unencrypted so that secure and unsecure traffic may flow in a single secure wire and inter-communicate. Rule 8 may be applied to permit secure and unsecure traffic to flow in a single secure wire, but isolated from each other. Rule 9 may be applied to encrypt all traffic over a secure wire.

In a further embodiment, the policy may specify that encryption is performed at a particular layer, such as at higher OSI layers (e.g., Layer 3 (L3) or Layer 4 (L4)). If L3 payloads are encrypted, for example, L3 devices such as routers may be able to see the frames' IP headers and route the frames. As a result, the traffic may be routed through devices which do not participate in the secure wire. If L4 payloads are encrypted, the traffic may be able to pass through network address translation (NAT) devices such as load balancers.

At step 550, the encryption module forwards the encrypted frame to the pNIC, and at step 560, the pNIC transmits the encrypted frame via a network. Note, the encrypted frame may be further modified before reaching the pNIC. For example, a module may add a VXLAN header to the encrypted frame in one embodiment. At step 570, a second pNIC which is connected to the same network receives the transmitted frame. The second pNIC may then forward the transmitted frame to dvSwitch (or other virtual switch), which is responsible for routing the frame to a second vNIC (vNIC 2) of a second VM (VM 2) based on a destination MAC address indicated in the transmitted frame. In one embodiment, the received frame may be modified before being forwarded to the dvSwitch. For example, in the case of VXLAN, a module may strip a VXLAN header from the received frame.

At step 580, a decryption module intercepts the encrypted frame as the frame is en route to the DVSwitch and decrypts the frame's data by invoking an API of a key management module of the receiving VM host. Here, the key management module may use an encryption key indicated by the Key ID field of the encrypted frame. As discussed, the key management module of the receiving VM host may lack the encryption key identified by the Key ID if vNIC 2 is not connected to the same secure wire as vNIC 1, in which case decryption will not succeed. If the encrypted frame includes fields such as a message digest field, the decryption module may further authenticate the sender, ensure data integrity, etc., as appropriate, based on the value of the message digest field.

At step 590, the decryption module forwards the decrypted frame to DVSwitch which in turn forwards it to vNIC 2. Here, the decrypted frame may be an ordinary Ethernet frame. Note, vNIC 2 does not see encrypted frames, and vNIC 1 sends normal Ethernet frames, so the secure wire may be transparent to both VM1 and VM2. In one embodiment, the decryption and encryption modules may be located at a layer between the virtual and physical switches. In such a case, the secure wire may further be transparent to the virtual switch, such as the DVSwitch.

FIG. 6 illustrates a method 600 for creating a secure wire and adding vNICs to the secure wire, according to an embodiment. As shown, the method 600 begins at step 610, where an infrastructure administrator creates an L2 domain and enables secure wire on the L2 domain. For example, the L2 domain and secure wire may be created on a network management platform (e.g., VMware Vshield™ Manager) which facilitates management of such domains for a multi-tenant environment (e.g., a cloud environment). The network management platform may then program an application running on a secure key management server to generate an encryption key for the secure wire. For example, a cloud infrastructure administration may create a VXLAN based virtual wire on the virtualization management platform and enable secure wire on the virtual wire. The virtualization management platform may then program an application running on the key management server to generate an encryption key for the new secure wire.

At step 620, a tenant administrator adds vNICs to the secure wire. Similar to the creation of the secure wire, discussed above, the tenant administrator may add vNICs to the secure wire on a virtualization management platform which manages such domains for a multi-tenant environment. In one embodiment, only tenant administrators are allowed to add vNICs to, or remove vNICs from, secure wires. Such an approach ensures that secure wire membership is controlled by tenants based on, for example, the need for secure communications between some of the tenants' VMs. Further, unauthorized VMs may be prevented from joining the secure wires.

At step 630, communication between vNICs on the secure wire is encrypted using the key associated with the secure wire. As discussed, a key manager application on the key management server may store secure wire encryption keys and membership information after secure wires are created and vNICs added to the secure wires, respectively. For a given secure wire, the key manager application may securely push the encryption key for the secure wire to key management modules of VM hosts which support vNICs added to the secure wire. Communication between the added vNICs may then be intercepted by encryption/decryption modules and encrypted/decrypted using the encryption key by invoking APIs exposed by the key management modules.

Although discussed above primarily with respect to multi-tenant cloud environments, techniques disclosed herein may also be used in other environments. For example, secure wires may provide protection and isolation to services running in enterprise environments. For example, secure wire may be used to isolate critical financial servers such that only finance department users have access to the isolated servers. Even if others obtain the data by trapping into the isolated network, the obtained data would be encrypted.

Although discussed above with respect to a secure wire which represents a single broadcast domain, multiple secure wires may interact via edge routers such as VMware vShield Edge®. Since the group encryption key is distinct for each secure wire, frames traversing from one secure wire to another cannot be decrypted by the latter. To solve this problem, an edge router may be configured such that one arm of the edge router belongs to a first secure wire and another arm of the edge router belongs to a second secure wire. The edge router may further decrypt frames traversing from the first secure wire to the second using the group key of the first secure wire, routed the frame, and encrypted the frame again using the group key of the second secure wire.

Advantageously, embodiments described herein secure data flowing over multi-tenant virtualized infrastructures using group key-based encryption which may be transparent to guest VMs and virtual switches. Even if malicious entities gain unauthorized access to secure wire data by means such as passive wiretapping, masquerading attacks, etc., the malicious entities will be unable to decipher the data. Further, group keying is well-suited for network encryption in virtualized and cloud environments, among others, due to group keying's elegant scalability, easy management, and ability to permit policies and keys to be controlled centrally. Group keying also eliminates the need to negotiate keys on a point-to-point basis, which may become intractable as the number of endpoints grows.

The various embodiments described herein may employ various computer-implemented operations involving data stored in computer systems. e.g., these operations may require physical manipulation of physical quantities usually, though not necessarily, these quantities may take the form of electrical or magnetic signals where they, or representations of them, are capable of being stored, transferred, combined, compared, or otherwise manipulated. Further, such manipulations are often referred to in terms, such as producing, identifying, determining, or comparing. Any operations described herein that form part of one or more embodiments of the invention may be useful machine operations. In addition, one or more embodiments of the invention also relate to a device or an apparatus for performing these operations. The apparatus may be specially constructed for specific required purposes, or it may be a general purpose computer selectively activated or configured by a computer program stored in the computer. In particular, various general purpose machines may be used with computer programs written in accordance with the teachings herein, or it may be more convenient to construct a more specialized apparatus to perform the required operations.

The various embodiments described herein may be practiced with other computer system configurations including hand-held devices, microprocessor systems, microprocessor-based or programmable consumer electronics, minicomputers, mainframe computers, and the like.

One or more embodiments of the present invention may be implemented as one or more computer programs or as one or more computer program modules embodied in one or more computer readable media. The term computer readable medium refers to any data storage device that can store data which can thereafter be input to a computer system. Computer readable media may be based on any existing or subsequently developed technology for embodying computer programs in a manner that enables them to be read by a computer. Examples of a computer readable medium include a hard drive, network attached storage (NAS), read-only memory, random-access memory (e.g., a flash memory device), a CD (Compact Discs), CD-ROM, a CD-R, or a CD-RW, a DVD (Digital Versatile Disc), a magnetic tape, and other optical and non-optical data storage devices. The computer readable medium can also be distributed over a network coupled computer system so that the computer readable code is stored and executed in a distributed fashion.

Although one or more embodiments of the present invention have been described in some detail for clarity of understanding, it will be apparent that certain changes and modifications may be made within the scope of the claims. Accordingly, the described embodiments are to be considered as illustrative and not restrictive, and the scope of the claims is not to be limited to details given herein, but may be modified within the scope and equivalents of the claims. In the claims, elements and/or steps do not imply any particular order of operation, unless explicitly stated in the claims.

In addition, while described virtualization methods have generally assumed that VMs present interfaces consistent with a particular hardware system, persons of ordinary skill in the art will recognize that the methods described may be used in conjunction with virtualizations that do not correspond directly to any particular hardware system. Virtualization systems in accordance with the various embodiments, implemented as hosted embodiments, non-hosted embodiments, or as embodiments that tend to blur distinctions between the two, are all envisioned. Furthermore, various virtualization operations may be wholly or partially implemented in hardware. For example, a hardware implementation may employ a look-up table for modification of storage access requests to secure non-disk data.

Many variations, modifications, additions, and improvements are possible, regardless the degree of virtualization. The virtualization software can therefore include components of a host, console, or guest operating system that performs virtualization functions. Plural instances may be provided for components, operations or structures described herein as a single instance. Finally, boundaries between various components, operations and data stores are somewhat arbitrary, and particular operations are illustrated in the context of specific illustrative configurations. Other allocations of functionality are envisioned and may fall within the scope of the invention(s). In general, structures and functionality presented as separate components in exemplary configurations may be implemented as a combined structure or component. Similarly, structures and functionality presented as a single component may be implemented as separate components. These and other variations, modifications, additions, and improvements may fall within the scope of the appended claims(s).

Claims

1. A computer-implemented method of providing group key-based encryption, comprising: receiving a selection of a Layer 2 (L2) domain on which a secure wire is to be enabled;generating, for the secure wire, an encryption key encrypting messages exchanged between a group of interfaces of a group of machines;associating, with the secure wire, a set of interfaces of a set of machines that execute on a set of host computers and that are part of the L2 domain, wherein each particular interface of each particular machine connects the particular machine to a virtual switch that executes on the particular machine's host computer;providing to the set of host computers the encryption key to encrypt, on each host computer in the set of host computers, frames that each particular machine of the set of machines passes along the particular machine's interface to another machine in the set of machines en route out of the particular machine's host computer; andproviding to the set of host computers a mapping between one or more secure wires for machines executing on the set of host computers and associated encryption keys.

2. The method of claim 1, wherein the encryption key is generated by a management server; andwherein the encryption key is securely transmitted from the management server to the one or more key management modules executing on the set of host computers.

3. The method of claim 1 further comprising: on each particular host computer in the set of computers: intercepting encrypted frames as the encrypted frames are received from outside of a particular host computer as the encrypted frames are en route to one or more destination machines executing on the particular host computer and associated with the secure wire; anddecrypting payload data of the encrypted frames using the encryption key.

4. The method of claim 1 further comprising: on each particular host computer in the set of computers, adding to each frame at least one of a value that identifies the encryption key and a signed hash value used to authenticate the encrypted frames and ensure data integrity.

5. The method of claim 4 further comprising adding to each frame an encoded encryption initialization vector value.

6. The method of claim 1, wherein the frames are encrypted according to IEEE MAC Security Standard (MAC Sec) frame format.

7. The method of claim 1 further comprising providing to each host computer in the set of host computers one or more policies specifying which traffic out of the set of machines to encrypt.

8. The method of claim 1, wherein the secure wire is enabled on a virtual extensible LAN (VXLAN).

9. The method of claim 1, wherein associating, with the secure wire, the set of interfaces of the set of machines comprises receiving input from an administrator to add the set of machines to the secure wire.

10. A non-transitory computer readable storage medium storing a program for execution by at least one processing unit, the program provides group key-based encryption, the program comprising sets of instructions for: receiving a selection of a Layer 2 (L2) domain on which a secure wire is to be enabled;generating, for the secure wire, an encryption key encrypting messages exchanged between a group of interfaces of a group of machines;associating, with the secure wire, a set of interfaces of a set of machines that execute on a set of host computers and that are part of the L2 domain, wherein each particular interface of each particular machine connects the particular machine to a virtual switch that executes on the particular machine's host computer;providing to the set of host computers the encryption key to encrypt, on each host computer in the set of host computers, frames that each particular machine of the set of machines passes along the particular machine's interface to another machine in the set of machines en route out of the particular machine's host computer; andproviding to the set of host computers a mapping between one or more secure wires for machines executing on the set of host computers and associated encryption keys.

11. The non-transitory computer readable storage medium of claim 10, wherein the encryption key is generated by a management server; andwherein the encryption key is securely transmitted from the management server to the one or more key management modules executing on the set of host computers.

12. The non-transitory computer readable storage medium of claim 10, wherein the program further comprises sets of instructions for: on each particular host computer in the set of computers: intercepting encrypted frames as the encrypted frames are received from outside of a particular host computer as the encrypted frames are en route to one or more destination machines executing on the particular host computer and associated with the secure wire; anddecrypting payload data of the encrypted frames using the encryption key.

13. The non-transitory computer readable storage medium of claim 10, wherein the program further comprises a set of instructions for: on each particular host computer in the set of computers, adding to each frame at least one of a value that identifies the encryption key and a signed hash value used to authenticate the encrypted frames and ensure data integrity.

14. The non-transitory computer readable storage medium of claim 13, wherein the program further comprises a set of instructions for adding to each frame an encoded encryption initialization vector value.

15. The non-transitory computer readable storage medium of claim 10, wherein the frames are encrypted according to IEEE MAC Security Standard (MAC Sec) frame format.

16. The non-transitory computer readable storage medium of claim 10, wherein the program further comprises a set of instructions for providing to each host computer in the set of host computers one or more policies specifying which traffic out of the set of machines to encrypt.

17. The non-transitory computer readable storage medium of claim 10, wherein the secure wire is enabled on a virtual extensible LAN (VXLAN).

18. The non-transitory computer readable storage medium of claim 10, wherein the set of instructions for associating, with the secure wire, the set of interfaces of the set of machines comprises a set of instructions for receiving input from an administrator to add the set of machines to the secure wire.