Multitenant access to multiple desktops on host machine partitions in a service provider network

BACKGROUND

0001It is common for an enterprise to use a service provider to host computer resources, such as servers, applications and desktops. Typically, the service provider has multiple computers, each of which is running one or more virtual machines. A virtual machine may host, for example, a server, or an application, or an entire desktop. An enterprise may have a service provider host many virtual machines.

0002In order to integrate a hosted desktop within an enterprise, the hosted desktop is assigned a network address that is consistent with the network addressing plan within the enterprise. In other words, the enterprise specifies the addresses of desktops that are hosted within a service provider; the service provider does not dictate the desktop addressing plan to the enterprise. To host an enterprise's desktops, the service provider communicates directly with agent software in the hosted desktop. In other words, the service provider uses the address space specified by the enterprise, and works within the enterprise's existing addressing plan.

0003Because of the Internet's limited address space, it is highly probably that large enterprises are using a private address space (as defined in RFC 1918) for internal addressing requirements. Further, it is highly probably that different enterprises are using the same private address ranges. (e.g., two enterprises may both address themselves using 192.168.0.0/16.). As a result, a service provider generally segregates resources for each enterprise that it hosts. In particular, each host machine is allocated statically to a single tenant.

0004<patcit id="pcit0001" dnum="US2011142053A"><text>US2011/142,053 (Van Der Merwe et al.</text></patcit>) describes coupling virtual private networks to virtual machines within a distributive computer network. The claims have been characterised over this document. <patcit id="pcit0002" dnum="US2011090911A"><text>US2011/090,911 (Hao et al.</text></patcit>) describes transparent cloud computing within a virtualised network infrastructure. <patcit id="pcit0003" dnum="US2008240122A"><text>US2008/240,122 (Richardson et al.</text></patcit>) describes configuring intercommunication between computing nodes. <patcit id="pcit0004" dnum="US2009276771A"><text>US2009/276771 (Nickolov Peter et al.</text></patcit>) describes a resource allocation management tool.

SUMMARY

0005According to the present invention there is provided a computer system as set forth in claim 1. Other features of the invention will be apparent from the dependent claims, and the description which follows.

0006Capital resources of a service provider can be more efficiently utilized by sharing some amount of infrastructure among enterprise tenants. For example, routers and host machines can be designed such that desktops can be hosted for two or more enterprises.

0007A virtual local area network is established for each enterprise. Each virtual local area network is connected to a plurality of host machines for the enterprise, with each host machine supporting desktops for use by the enterprise. The desktops in turn access the computer resources on the enterprise network of the enterprise to which it is connected.

0008To eliminate network conflicts, a router is connected to the enterprise networks of the multiple enterprises. A virtual router is established for each enterprise, including a distinct routing table for each enterprise, for routing traffic between the desktops hosted on the plurality of host machines and the computer resources on the enterprise networks.

0009As an example, a router can tag network addresses in packets received by the router based on a physical, or virtual port on the router on which the packet is received and the enterprise assigned to that physical port. A separate virtual routing and forwarding (VRF) table can be created for each enterprise. The router connects to a virtual local area network (VLAN) for the enterprise, with different enterprises having their network addresses tagged differently. In effect, the router is virtualized. AVRF-enabled router can be used for such an implementation.

0010With such an implementation, host machines can be reallocated dynamically to different enterprises by changing the network address of the host machine to include the tag for the corresponding enterprise.

0011To allow resources within the host machine to be shared, a host machine includes a port group within a virtual switch for each enterprise it supports. The virtual switch for an enterprise is connected to the virtual local area network of the enterprise. Desktops in the host machine that are allocated to the enterprise are given network addresses that include the tag for that enterprise. The virtual switch ensures that only packets intended for an enterprise's desktop are seen only in that enterprise's network.

0012With this architecture, virtual desktops for different enterprises can be hosted on different partitions of the same host machine, thus allowing the resources on one machine to be shared among multiple tenants of a service provider.

0013When a host is partitioned in this manner, and shared among multiple tenants, the usage of the host by one tenant could affect the performance of the host experienced by another tenant. To mitigate the impact of one tenant's usage on the performance of the host for other tenants, each tenant can have its desktops pinned to one or more physical CPU's, separate from the physical CPU's used by other tenants. An alternative way to address this potential problem is to allocate shares of CPU resources to each virtual machine or to dynamically control bursting, or sudden increase, of CPU usage of individual machines, thus providing greater CPU resources to a desktop than may ordinarily be possible. A resource manager that is monitoring the resources of each virtual machine on a host can control the throttling of the CPU to ensure an acceptable quality of service to all desktops for all enterprises that may be sharing a host.

BRIEF DESCRIPTION OF THE DRAWINGS

0014<ul id="ul0001" list-style="none" compact="compact"><li><figref idref="f0001">Fig. 1</figref> is a block diagram of a service provider infrastructure with host machines supporting multiple virtual machines for multiple enterprises.</li><li><figref idref="f0002">Fig. 2</figref> is an illustration of an example desktop model quota data structure.</li><li><figref idref="f0002">Fig. 3</figref> is an illustration of data associated with an enterprise.</li><li><figref idref="f0003">Fig. 4</figref> is a flow chart describing an example process through which available hosts are identified.</li></ul>

DETAILED DESCRIPTION

0015Referring now to <figref idref="f0001">Fig. 1</figref>, a block diagram of a service provider infrastructure with host machines supporting multiple virtual machines for multiple enterprises will now be described.

0016A first enterprise, such as a corporation, has its own computer network 100 through which computers 102 and other resources of the enterprise communicate. The computer network 100 can be any combination of local area networks and wide area networks.

0017A second enterprise has its own computer network 180 through which computers 182 and other resources of the second enterprise communicate. The computer network 180 can be any combination of local area networks and wide area networks.

0018Through one or more routers 150, the computer networks 100, 180 are connected to a computer network 152, on which multiple virtual local area networks (VLANs) are implemented.

0019To eliminate network conflicts, for each router that is connected to multiple enterprise networks 100, 180, a virtual router is established for each enterprise. The virtual router includes a distinct routing table for each enterprise. As an example, a router can tag network addresses in packets received by the router based on a physical or virtual port on the router on which the packet is received and the enterprise assigned to that port. A separate virtual routing and forwarding (VRF) table can be created for each enterprise. The router connects to the virtual local area network (VLAN) for the enterprise, with different enterprises having their network addresses tagged differently. In effect, the router is virtualized. AVRF-enabled router can be used for such an implementation. For example, a Multiprotocol Label Switching (MPLS)-enabled router can be used, in which case physical ports are assigned to enterprises. As another example, a router supporting IPSec tunnels, with an IPSec tunnel assigned to each enterprise can be used, in which case the ports assigned to an enterprise are virtual.

0020In an example implementation, the router tags network addresses in packets received by the router from an enterprise based on a physical or virtual port on the router on which the packet is received and the enterprise assigned to that port. The packets with tagged addresses are output from the router to the virtual local area network for that enterprise. Similarly, when the router receives a packet with a tagged address from the virtual local area network for an enterprise, it strips the tagging and outputs the packet on the port assigned to the enterprise corresponding to the tag.

0021Through a switching layer 154, the VLANs connect to multiple host computers 120. Each of the host computers 120 can run one or more virtual machines 122. A virtual machine 122 can run, for example, a desktop, and manage the interaction between a remote computer on an enterprise network and the desktop. A virtual machine also can run an application or a server.

0022To allow resources within the host machine to be shared, a host machine includes a port group within a virtual switch for each enterprise it supports. The virtual switch for an enterprise is connected to the virtual local area network of the enterprise. Desktops in the host machine that are allocated to the enterprise are given network addresses that include the tag for that enterprise and are associated with the virtual switch for that enterprise. The virtual switch ensures that only packets intended for an enterprise's desktop are seen only in that enterprise's network.

0023The service provider infrastructure also includes a management system 130 connected to host computers 120, separately from the enterprise networks through a backbone network 140. The management system 130 can include one or more computers. A service provider management tool 132 is used to allocate infrastructure capacity to an enterprise. A tenant management tool 134 is used to allocate desktop capacity. The service provider management tool is not connected to the enterprises' virtual area networks, but instead connects to host machines or other resources through a separate network, the backbone network 140. Thus, the service provider does not have access to an enterprise's resources or desktops on the host machines. The tenant management tool, on the other hand has access to the virtual area network of the enterprise, to allocate desktops within the quotas established with the service provider.

0024As an example implementation of the system illustrated in <figref idref="f0001">Fig. 1</figref>, some data structures for tracking host usage will now be described.

0025A desktop model quota data structure shown in <figref idref="f0002">Fig. 2</figref> can be used to track information about allocated resources. As an example, this data structure can include information such as: a maximum number 200 of central processing units (CPUs) supporting virtual machines (VMs), a maximum amount 202 of memory for each virtual machine, the current number 204 of CPUs used, a current amount 206 of memory used, and host identifiers 208 that can be assigned to VMs when desktops are provisioned on them.

0026As shown in <figref idref="f0002">Fig. 3</figref>, the service provider associates this desktop model quota 306 with an enterprise, which also has an organization identifier 300 and a virtual local area network (VLAN) identifier 302 (which is stored as part of the host configuration information). Thus, the combination of the organization identifier 302, virtual local area network identifier 308 and host identifiers relates all of the resources together for an enterprise. A desktop model 304, defining parameters of the desktops to be allocated, also is associated with an organization identifier and thus an enterprise.

0027To determine the host identifiers that can be associated with a desktop model quota, a process for retrieving the available hosts is implemented. This process receives the organization identifier and the desktop identifier and generates a list of available hosts that are capable of supporting the identified desktop model.

0028The process of identifying available hosts can cycle through each host in the system that is capable of being connected to the enterprise's VLAN. For example, referring to <figref idref="f0003">Fig. 4</figref>, for each selected (400) host, it is determined 402 how many desktops it can support, if any. These steps are repeated for each host until enough host capacity has been identified to support the desktop model quota, as indicated at 404. The identified hosts are then associated 406 with the desktop model quota.

0029To associate each virtual machine on a host with the enterprise, a VLAN name for the enterprise's VLAN is identified. The network label for the network adapter associated with the virtual machine is set to the VLAN name for the enterprise to which it is allocated. Thus all VM's belonging to the same tenant in a data center carry the same label. The VLAN name can be used to retrieve the VLAN identifier which is available from the host configuration information.

0030In addition to the foregoing, each host can have a data structure that maintains information to allow it to track usage and capacity metrics. The host data structure includes, but is not limited to, a marker indicating whether the host is shared (partitioned) or not, an indication of a memory over-allocation rate used in calculating the memory capacity of the host, a VM ratio indicating the ratio of virtual to physical machines (used in calculating the CPU capacity of the host), the CPU capacity of the host (calculated by the product of the total number of CPUs and the VM ratio), and the memory capacity of the host (calculated by the product of the total memory of the host by the memory over-allocation rate).

0031Given this infrastructure, additional methods can be provided for determining the current usage of a host or of a tenant. For example, information about the currently active virtual machines on each host can be retrieved. These can then be filtered by their VLAN identifier to segregate them by tenant. The methods can be implemented in the management system.

0032With this architecture, virtual desktops for different enterprises can be hosted on different partitions of the same host machine, thus allowing the resources on one machine to be shared among multiple tenants of a service provider.

0033When a host is partitioned in this manner, and shared among multiple tenants, the usage of the host by one tenant could affect the performance of the host experienced by another tenant. To mitigate the impact of one tenant's usage on the performance of the host for other tenants, each tenant can have its desktops pinned to one or more physical CPU's, separate from the physical CPU's used by other tenants. An alternative way to address this potential problem is to allocate shares of CPU resources to each virtual machine or to dynamically control bursting of CPU usage of individual machines, thus providing greater CPU resources to a desktop than may ordinarily be possible. A resource manager that is monitoring the resources of each virtual machine on a host can control the throttling of the CPU to ensure an acceptable quality of service to all desktops for all enterprises that may be sharing a host.

0034The techniques described above can be implemented in digital electronic circuitry, or in computer hardware, firmware, software executing on a computer, or in combinations of them. The techniques can be implemented as a computer program product, i.e., a computer program tangibly embodied in tangible, machine-readable storage medium, for execution by, or to control the operation of, data processing apparatus, e.g., a programmable processor, a computer, or multiple computers. A computer program can be written in any form of programming language, including compiled or interpreted languages, and it can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. A computer program can be deployed to be executed on one computer or on multiple computers at one site or distributed across multiple sites and interconnected by a communication network.

0035Method steps of the techniques described herein can be performed by one or more programmable processors executing a computer program to perform functions described herein by operating on input data and generating output. Method steps can also be performed by, and apparatus of the invention can be implemented as, special purpose logic circuitry, e.g., an FPGA (field programmable gate array) or an ASIC (application-specific integrated circuit). Applications can refer to portions of the computer program and/or the processor or other circuitry that implements that functionality.

0036Processors suitable for the execution of a computer program include, by way of example, both general and special purpose microprocessors, and any one or more processors of any kind of digital computer. Generally, a processor will receive instructions and data from a read-only memory or a random access memory or both. The essential elements of a computer are a processor for executing instructions and one or more memory devices for storing instructions and data. Generally, a computer will also include, or be operatively coupled to receive data from or transfer data to, or both, one or more mass storage devices for storing data, e.g., magnetic, magneto-optical disks, or optical disks. Storage media suitable for embodying computer program instructions and data include all forms of non-volatile memory, including by way of example semiconductor memory devices, e.g., EPROM, EEPROM, and flash memory devices; magnetic disks, e.g., internal hard disks or removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks. The processor and the memory can be supplemented by, or incorporated in special purpose logic circuitry.

0037A computing system can include clients and servers. A client and server are generally remote from each other and typically interact over a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.

0038Having described an example embodiment, it should be apparent to those skilled in the art that the foregoing is merely illustrative and not limiting, having been presented by way of example only. Numerous modifications and other embodiments are with the scope of ordinary skill in the art and are contemplated as falling with the scope of the invention. as defined by the appended claims.