[0001] The present subject matter relates to cloud computing networks and,
particularly but not exclusively, to managing flow migration between virtual network
appliances in the cloud computing network.
BACKGROUND
[0002] Cloud computing networks have reshaped the field of Internet-provided
services due to its beneficial nature for individual users as well as large enterprises. The cloud
computing networks utilize virtual machines (VMs) for providing various services, such as
firewalls, data storage, and intrusion detection to users. The VM may be understood as a
portion of software that, when executed, allows virtualization of an actual physical computing
system. Each VM may function as a self-contained platform, running its own operating
system and software applications. Cloud computing customers are thus able to access various
services and applications without actually purchasing physical resources utilized for the
services.
SUMMARY
[0003] This summary is provided to introduce concepts related to systems and
methods for flow migration between virtual network appliances in a cloud computing
network. This summary is neither intended to identify essential features of the claimed
subject matter nor is it intended for use in determining or limiting the scope of the claimed
subject matter.
[0004] In one implementation, a network appliances managing architecture for
migrating flow between virtual network appliances (VNA) is described. The network
appliances managing architecture comprises a controller to receive performance data for a
VNA and analyze the performance data to determine whether the VNA has a weak
performance status. The weak performance status corresponds to any one of an overloaded,
an under-loaded, and a failed status. The network appliances managing architecture further
comprises a classifier to receive a flow migration request from the controller for migrating
one or more flows of data packets from the VNA based on the analyzing. The classifier
3
further identifies an active VNA for flow migration based on a predetermined mapping policy
and migrates the one or more flows from the VNA to the at least one active VNA.
[0005] In another implementation, a method for flow migration in a cloud computing
network is described. The method includes receiving performance data for a VNA. The
method further comprises analyzing the performance data to determine whether the VNA has
a weak performance status, where the weak performance status corresponds to any one of an
overloaded, an under-loaded, and a failed status. Further, a flow migration request is provided
to a classifier for migrating one or more flows of data packets from the VNA based on the
analyzing. Further at least one active VNA is identified for flow migration based on a
predetermined mapping policy. The method further comprises migrating the one or more
flows from the VNA to the at least one active VNA.
[0006] In yet another implementation, a method for managing virtual network
appliances (VNAs) is described. The method for managing the VNAs comprises ascertaining
total load handled by a plurality of VNAs operating in a cloud computing network. Further,
the total load is compared with a minimum threshold level and a maximum threshold level.
The method further comprises determining whether to perform at least one of a scaling up or
scaling down of the plurality of VNAs based on the comparing. Further, at least one VNA is
identified from among the plurality of VNAs for flow migration based on the determination.
The method further comprises providing a flow migration request to a classifier for migrating
one or more flows of data packets from the at least one VNA based on the identifying. The
method further comprises migrating the one or more flows from the at least one VNA to at
least one active VNA based on a predetermined mapping policy.
[0007] In yet another implementation, a computer-readable medium having embodied
thereon a computer program for executing a method of flow migration between virtual
network appliances (VNAs) in a cloud computing network is described. The method
comprises receiving performance data for a VNA. The method further comprises analyzing
the performance data to determine whether the VNA has a weak performance status, where
the weak performance status corresponds to any one of an overloaded, an under-loaded, and a
failed status. Further, a flow migration request is provided to a classifier for migrating one or
more flows of data packets from the VNA based on the analyzing. Further at least one active
VNA is identified for flow migration based on a predetermined mapping policy. The method
4
further comprises migrating the one or more flows from the VNA to the at least one active
VNA.
BRIEF DESCRIPTION OF THE FIGURES
[0008] The detailed description is described with reference to the accompanying
figures. In the figures, the left-most digit(s) of a reference number identifies the figure in
which the reference number first appears. The same numbers are used throughout the figures
to reference like features and components. Some embodiments of system or methods in
accordance with embodiments of the present subject matter are now described, by way of
example only, and with reference to the accompanying figures, in which:
[0009] Figure 1 illustrates a cloud computing environment, according to an
embodiment of the present subject matter.
[0010] Figure 2 illustrates a method for flow migration between virtual network
appliances in a cloud computing network, according to an embodiment of the present subject
matter.
[0011] Figure 3 illustrates a method for managing virtual network appliances,
according to an embodiment of the present subject matter.
[0012] It should be appreciated by those skilled in the art that any block diagrams
herein represent conceptual views of illustrative systems embodying the principles of the
present subject matter. Similarly, it will be appreciated that any flow charts, flow diagrams,
state transition diagrams, pseudo code, and the like, represent various processes which may
be substantially represented in computer readable medium and so executed by a computer or
processor, whether or not such computer or processor is explicitly shown.
DESCRIPTION OF EMBODIMENTS
[0013] Systems and methods for flow migration between virtual network appliances
in a cloud computing network are described. Cloud computing is a conventionally known
technique of providing services to users by way of creating virtual environment of computing
resources. The cloud computing network involves hardware and software resources,
accessible through Virtual Machines (VM), hosted on either the Internet or a private network
to form a virtual environment for providing various services, such as firewalls, data storage,
5
WAN optimization, and intrusion detection. The VMs, as will be understood, are computing
machines with a software that when executed create a replica of a physical machine for
providing same services that are provided by the physical machine but in a virtual
environment. Thus, VMs are typically used to virtualize computing machines and network
appliances, such as end user applications, firewalls, data storage devices, WAN optimizers,
virtual private networks (VPNs), and intrusion detection system. The VMs function as
independent machines running its own operating system, processors, and other software
applications. Any user or subscriber may thus subscribe with the service provider who is
providing a cloud computing network service and may interact with the VMs for using the
service.
[0014] Typically, the service providers create one or more virtual network appliances
(VNAs) corresponding to the physical network appliances in order to cater to a large number
of customers. The VNAs may be understood as the different virtual machines having its own
processor(s) running its own operating system, and other software applications independently
of each other. For instance, a service provider hosting a cloud computing network for
providing virtual firewalls may host various VNAs with each VNA acting as an independent
firewall for a particular set of users. The users may thus interact with any of the VNAs
providing a similar service for availing the particular service. For the purpose, the
conventional cloud computing networks include a load balancer to divert flow of data packets
from the user to the VNAs based on various factors, such as load, i.e., number of flows
handled by each of the VNA.
[0015] Typically, upon receiving a new flow of data packets, hereinafter referred to as
flow, the load balancer may determine the load handled by each of the VNAs and direct the
flow to the VNA having least load. In case all the VNAs are heavily loaded, the load balancer
may launch a new VNA and map the flow to the new VNA, thus managing the load in the
cloud computing network. Although launching the new VNA may facilitate the load balancer
in directing the newly received flows, the load balancer still may not be able to reduce the
load on the existing VNAs as flows, once mapped to a VNA, have to be typically managed
and processed by the same VNA. The new VNA may thus be able to handle only the new
flows and may thus not be launched for very few flows as each VNA may result in additional
costs for the service provider. The load balancer may thus launch the new VNAs only when it
either receives lot of new flows or when the existing VNAs are very heavily loaded, thus
6
affecting the efficiency of the existing VNAs due to the limited resources, such as processor
capabilities and memory space.
[0016] Further, in order to manage resource utilization and for reducing the associated
costs, the conventional load balancers may reduce the number of VNAs whenever the load
reduces. For the purpose, the load balancers may remove the VNAs handling least number of
flows. However, as a VNA may be removed after all the flows managed by the VNA are
processed, the load balancer may not be able to remove the VNAs immediately, thus resulting
in resource and wastage. Removing a VNA may be difficult especially in services that
involve long continuous flows, for instance, in cases of VPN connections. Furthermore, in
case any VNA fails, i.e., stops functioning, the flow managed by the VNA may get disrupted,
thus affecting quality of the service offered by the service provider and received by the
customer. The load balancer may thus either wait for the VNA to restart functioning or map
the flow to another VNA for restarting the flow processing from the beginning, thus affecting
customer’s experience, especially when the failed VNA had been processing the flow for a
long time.
[0017] According to an implementation of the present subject matter, systems and
methods for flow migration between virtual network appliances in a cloud computing
network are described. The systems and the methods can be implemented by a variety of
computing devices hosting virtual machines, such as a desktop computer, cloud servers,
mainframe computers, workstation, a multiprocessor system, a network computer, and a
server. Further, the systems and methods may be implemented in cloud computing networks
hosting variety of services, such as firewalls, data storage, WAN optimization, VPN,
intrusion detection, and data storage.
[0018] In accordance with an embodiment of the present subject matter, a network
appliances managing architecture for migrating flow between the VNAs in the cloud
computing network is described. In said embodiment, the network appliances managing
architecture may manage the flow between various VNAs such that flow from a first VNA
can be migrated to a second VNA in case the first VNA becomes overloaded, under-loaded,
or fails to operate. Migrating the flow allows the cloud computing network to efficiently
manage and process the flows without disrupting processing of the flow and in a cost
effective way. The network appliances managing architecture includes a controller, a
classifier, the VNAs, and global state database.
7
[0019] In one implementation, the classifier may map flows to the VNAs based on
one or more predetermined mapping policy. On receiving first packets of any new flow, the
controller may identify the VNA based on the mapping policy and instruct the VNA to
manage and process the flow. On receiving the instructions from the classifier, the VNA may
start receiving the data packets and initiate the processing of the flow. Further, the VNA may
include an agent for maintaining a local state of all the flows being processed by the VNA in
order to indicate the progress of the flow processing. In one implementation, the agent may
update the local state of the flow to the global state database for maintaining a progress report
of all the flows being processed in the cloud computing network. Maintaining such a report
allows easy and efficient migration of the flows between the VNA as a new VNA may easily
access the global state database to obtain the state of the flow processing and continue
processing the flow without affecting the processing of the flow.
[0020] Further, the agent may monitor resource utilization of the VNA and regularly
provide performance data indicating the resource utilization to the controller. The controller,
on receiving the performance data, may analyze the performance data to determine if the
VNA has a weak performance status, i.e., if the VNA is under-loaded, overloaded, or has
failed. In case the VNA is ascertained to have the weak performance status, the controller
may indicate the classifier to migrate the flows handled by the VNA to another VNA. The
controller may simultaneously also request the VNA to update its local state to the global
state database. The classifier may subsequently identify one or more active VNAs, i.e., VNAs
that are operational and are not overloaded and map the flows to one or more than one VNA.
In one implementation, the controller may also launch a new VNA in case the existing VNAs
may not be able to handle the load, for instance, in case of failure or overloading of one or
more VNAs. The classifier, in such a case, may identify the new VNA as the active VNA for
flow migration. The active VNA may subsequently access the global state database to obtain
global state of the migrated flow and start processing the flow.
[0021] Further, in one embodiment, the controller may facilitate fast scaling up and
scaling down of the VNAs in the cloud computing network in order to manage the resource
utilization of cloud computing network. For the purpose, the controller may analyze the
performance data of all the VNAs and determine the VNAs for which the flows may be
migrated to another VNA for efficient management and subsequently instruct the classifier to
migrate the flow from the VNA to the other VNA.
8
[0022] The present subject matter thus facilitates efficient and immediate migration of
flows between various VNAs in a cloud computing network. Providing the agents in the
VNA for continuously monitoring the local state of the VNA’s flow processing and
periodically update the global state database about global state of the VNA helps in keeping a
track of the progress of the flows being processed by the VNA. Thus, after flow migration,
the new VNA may easily continue processing of the flow from the same point at which the
flow was migrated from the earlier VNA. Further, enabling the controller to identify the
VNAs whose flows need to be migrated facilitates in ensuring smooth and efficient
functioning of the cloud computing network as all flows are continuously processed without
any interruption even when any VNA fails.
[0023] It should be noted that the description and figures merely illustrate the
principles of the present subject matter. It will thus be appreciated that those skilled in the art
will be able to devise various arrangements that, although not explicitly described or shown
herein, embody the principles of the present subject matter and are included within its spirit
and scope. Furthermore, all examples recited herein are principally intended expressly to be
only for pedagogical purposes to aid the reader in understanding the principles of the present
subject matter and the concepts contributed by the inventor(s) to furthering the art, and are to
be construed as being without limitation to such specifically recited examples and conditions.
Moreover, all statements herein reciting principles, aspects, and embodiments of the present
subject matter, as well as specific examples thereof, are intended to encompass equivalents
thereof.
[0024] It will also be appreciated by those skilled in the art that the words during,
while, and when as used herein are not exact terms that mean an action takes place instantly
upon an initiating action but that there may be some small but reasonable delay, such as a
propagation delay, between the initial action and the reaction that is initiated by the initial
action. Additionally, the words “connected” and “coupled” are used throughout for clarity of
the description and can include either a direct connection or an indirect connection.
[0025] The manner in which the systems and the methods of migrating flow between
virtual network appliances in the cloud computing network may be implemented has been
explained in details with respect to the Figures 1 to 3. While aspects of described systems and
methods for managing virtual network appliances in the cloud computing network can be
9
implemented in any number of different computing systems and transmission environments,
the embodiments are described in the context of the following system(s).
[0026] Figure. 1 illustrates a cloud computing environment 100 according to an
embodiment of the present subject matter. The cloud computing environment 100 includes
one or more user devices 102-1, 102-2 , 102-3, …., 102-n, hereinafter collectively referred to
as user devices 102 and individually referred to as user device 102, communicating with a
cloud computing network 104 for accessing one or more services offered by a host of the
cloud computing network 104. In one implementation, the user device 102 may communicate
with the cloud computing network 104 over one or more communication links.
[0027] In one implementation, the cloud computing network may provide a variety of
services, such as firewalls, data storage, WAN optimization, VPN, intrusion detection, and
data storage. A service provider hosting the cloud computing network 104, hereinafter
referred to as cloud 104, may install one or more of a variety of computing devices (not
shown in the figure), such as a desktop computer, cloud servers, mainframe computers,
workstation, a multiprocessor system, a network computer, and a server for hosting one or
more virtual machines for offering the variety of services to the user device 102. In one
implementation, the computing devices may host individual virtual machines for each of the
services hosted by the cloud 104. Further, in order to serve a large number of the user devices
102, the computing devices may host one or more virtual network appliances (VNAs) 106-1,
106-2, …., 106-N, hereinafter referred to as VNAs 106 of the same network appliance in the
cloud 104. Each of the VNAs 106 may thus be understood as a self-contained platform
having its own processers and memory spaces for running its own operating system and
software applications.
[0028] For instance, in case of a cloud providing firewall services, the cloud 104 may
include one or more VNAs 106 for providing firewall services to the user devices 102 such
that each of the VNAs 106 serve one or more of the user devices 102. In order to avail the
services offered by the cloud 104, each of the user devices 102 may exchange data packets
with the cloud 104. On receiving the data packets, the VNAs 106 may process the data
packets for providing the services to the user devices 102. In one implementation, the cloud
104 may monitor the distribution of flows of data packets among the VNAs 106 such that one
10
or more flows handled by a VNA 106 may be efficiently and immediately migrated to
another VNAs 106 as and when desired for managing the load handled by the VNAs 106.
[0029] For the purpose, the cloud 104 may implement a network appliances managing
architecture 108 comprising a flow distribution system 110, the VNAs 106, and global state
database 112. The network appliances managing architecture 108 may facilitate the cloud 104
in the efficient migration of the flows among the VNAs 106. In one implementation, the flow
distribution system 110 may be a virtual machine to distribute the flow received from the user
devices 102 among the VNAs 106. The flow distribution system 110 may further include a
classifier 114 to distribute the flows among the VNAs 106 and a controller 116 to manage
migration of the flows among the VNAs 106. Although the flow distribution system 110 has
been shown as a single system residing on a single virtual machine, it may be implemented as
a distributed system with the controller 116 and the classifier 114 residing as separate virtual
machines. Further, the controller 116 and classifier 114 may, individually or collectively,
reside on any of the VNAs 106.
[0030] The global state database 112 is provided to maintain a global state of the
VNAs 106. The global state of a VNA may be understood as data indicating progress of
processing of all the flows handled by the VNA. In one implementation, the global state may
further include rules applicable for processing the flows handled by the VNAs. The global
state database 112 may thus interact with the VNAs 106 over a data channel for exchanging
the global state data. Further, the global state database 112 may be maintained using known
techniques, such as distributed hash tables. Although the global state database 112 has been
shown as a single database residing on a single virtual machine, it may be implemented as a
distributed database residing on separate virtual machines, such as the VNAs 106.
[0031] During operation, a user of the user devices 102 intending to avail the services
provided by the cloud 104 may initially access a web based link provided by the service
provider to establish a communication channel with the cloud 104 using a communication
link, such as Internet. Once connected, the cloud 104 may start receiving data packets from
the user device 102 over the communication channel. In one implementation, the user devices
102 may use known TCP-IP protocols for interacting with the cloud 104 by transmitting data
packets defined by TCP flows. In one implementation, the data packets of every new flow
coming to the cloud 104 may be initially received by the classifier 114. On receiving the data
11
packets, the classifier 114 may identify a VNA 106 from among the VNAs 106 for handling
the new flow based on one or more predetermined mapping policies, such as round-robin
policy; policies based on load, such as memory and processor utilization; policies based on
flow type, such as http and ftp; and policies based on source and destination address of the
flow.
[0032] Upon identification, the classifier 114 may send a set of flow mapping
instructions to the identified VNA 106 for initiating handling of the new flow of data packets.
On receiving the instructions from the classifier 114, the VNA 106 may start receiving the
data packets and initiate the processing of the flow. For instance, in an example of the cloud
104 providing WAN optimization service the VNAs 106, implemented as WAN optimizers,
may perform fingerprint or SHA-hash computation of the data packets for redundancy
elimination functionality. In another example of the cloud 104 providing hosting the VNAs
106 as VPN servers, the VNAs 106 may process the data packets for performing encryption
to ensure secure transmission of the data packets. Further, the VNA 106 may interact with the
global state database 112 to obtain the global state having the rules for processing the flow.
[0033] For the purpose, each of the VNAs 106 may include an agent 118-1, .., 118-n,
hereinafter collectively referred to as the agents 118 and individually referred to as the agent
118, to obtain the global state from the global state database 112 based on which the VNA
106 may process the data packets of the flow. Further, the agent 118 may maintain a local
state of all the flows being processed by the VNA 106. The local state may be understood as
data, such as number of data packets received and processed by the VNA 106 thus indicating
progress of processing of the flow. In one implementation, the agents 118 may save the local
state in the local state data 120-1, …, 120-n, hereinafter collectively referred to as the local
state data 120 and individually referred to as the local state data 120. The agent 118 may thus
continuously monitor the progress of the flow and update the local state data 120 about the
local state.
[0034] The agent 118 may further update the local state of the flow to the global state
database 112 on periodic bases, for instance, upon processing of a predetermined number of
data packets or on regular intervals of time. In one implementation, upon each of such
updates, the agent 118 may reset the local state to Null and restart the monitoring of the local
state so that local state data 120 may only store the local state between two updates thus
utilizing very less amount of memory space. In another implementation, the agent 118 may
continue monitoring of the local state from the state at the time of the update.
12
[0035] In one implementation, the agent 118 may maintain performance data for the
VNA 106, where the performance data may include values performance parameters of the
VNA 106. Examples of the performance parameters include, but are not limited to, processor
utilization, memory utilization, and number of flows handled by the VNA 106. The
performance parameters may thus indicate the capabilities of the VNA 106, i.e., the amount
of load currently handled by the VNA 106 and the amount of load it can handle. The agent
118 may thus regularly monitor the performance parameter in order to monitor the health
status of the VNA 106 and provide the performance data to the controller 116. In one
embodiment, the agent 118 may provide the performance data upon receiving a request from
the controller 116.
[0036] On receiving the performance data, the controller 116 may analyze the
performance data to determine the performance status of the VNA 106. In one
implementation, the controller 116 may analyze the performance parameters to classify the
VNA 106 as having a performance status from one of under-loaded, overloaded, failed, and
balanced loaded. The under-loaded performance status may indicate that the load, i.e.,
number of data packets handled by the VNA 106 is less than a predetermined threshold of
minimum load, indicating that the VNA 106 is underperforming and can thus be either
removed or provided more load. The overloaded status may indicate that the number of data
packets handled by the VNA 106 is more than a predetermined threshold of maximum load
indicating that the VNA 106 is handling more load than its capability and thus needs to be
offloaded. The balanced loaded status may indicate that the number of data packets handled
by the VNA 106 is between the predetermined threshold of minimum load and the
predetermined threshold of maximum load. The failed status may indicate that the VNA 106
has malfunctioned and thus cannot handle the flows mapped to the VNA 106. In case the
controller 116 determines the VNA 106 to have a weak performance status, i.e., if the VNA
106 is under-loaded, overloaded, or has failed, the controller 116 may identify the VNA 106
as a weak VNA 106. The controller 116 may subsequently provide a flow migration request
to the classifier 114 for migrating one or more flows of the weak VNA 106 to another VNA
106. The controller 116 may simultaneously also instruct the weak VNA 106 to update its
local state to the global state database 112. In one implementation, the controller 116 may
send the flow migration request to the classifier 114 and the instructions to the VNA 106 over
a control channel. In one embodiment, in case the controller 116 determines that the existing
VNAs 106 may not be able to handle the flows being migrated from weak VNA 106, the
13
controller 116 may also launch a new VNA and inform the classifier 114 accordingly. For
instance, in case of failure or overloading of one or more VNAs 106, the controller 116 may
determine that the existing VNAs 106 may not be able to handle the migrated flows and may
thus launch a new VNA. In another embodiment, the controller 116 may launch a new VNA
whenever an existing VNA 106 fails.
[0037] For example, in case where each of the VNAs 106 is capable of handling 100
data packets per second, and total of three VNAs 106 are active, then the maximum load that
can be handled by the cloud will be 300 data packets per second. Let’s assume each of the
three VNAs 106 is handling flows having 75 data packets per second in total and one of the
VNA 106 fails. In such a case the controller 116 may determine that the flows of the weak
VNA 106 may not be handled by the other two VNAs 106 and may thus launch a new VNA.
The controller 116 may then send the flow migration request to the classifier 114 and also
inform the classifier 116 about the launching of the new VNA.
[0038] On receiving the flow migration request the classifier 114 may identify at least
one active VNA, e.g., a VNA 106 that is operational and is not overloaded and may thus be
able to handle one or more flows migrated from the weak VNA 106. In one implementation,
the classifier 114 may identify the VNA 106 based on one or more predetermined factors,
such as the performance status and the number of flows or data packets handled by the weak
VNA 106 and the other VNAs 106 currently operational in the cloud 104. For the purpose,
the classifier 114 may analyze the flow migration request to determine the number of flows
and data packets handled by the weak VNA 106 and whether the weak VNA 106 is underloaded,
overloaded, or failed. In case the VNA 106 is under-loaded, the classifier 114 may
ascertain that the controller 116 would not have launched any new VNA 106. The classifier
114 may thus identify at least one VNA 106 from the plurality of the VNAs 106, having the
performance status as either under-loaded or balanced loaded as the active VNA 106.
[0039] In case the VNA 106 is either over-loaded or failed, the classifier 114 would
ascertain that the controller 116 may have launched a new VNA. In case the new VNA has
been launched, the classifier 114 may identify the new VNA as the active VNA 106,
otherwise, the classifier 114 may identify at least on VNA 106, from the plurality of the
VNAs 106, having the performance status as either under-loaded or balanced loaded as the
active VNA 106. Further, based on the number of flows to be migrated, the classifier 114
may determine the number of active VNAs 106 to whom the flow needs to be migrated.
14
[0040] Upon identifying the new VNA 106, the classifier 114 may migrate the flows
from the weak VNA 106 to the at least one active VNA 106 by mapping the flows to the at
least one active VNA 106. Further, in case the weak VNA 106 is under-loaded, the classifier
114 may remove the weak VNA 106 upon flow migration. The classifier 114 may
subsequently send flow mapping instructions to the active VNA 106 instructing the active
VNA 106 to start processing the migrated flows. On receiving the flow mapping instructions,
the agent 118 of the active VNA 106 may subsequently access the global state database 112
to obtain the global state of the migrated flow and store in the local state data 120. Based on
the global state, the active VNA 106 may determine the rules for processing the flow and
processing status of the migrated flow and subsequently start processing the flow without
affecting the flow process. Storing the global state in the global state database 112 thus
facilitates the active VNA 106 in efficiently processing the flow from the same point where
the weak VNA 106 had stopped processing the flow, thus saving substantial resources.
Further, saving the global state in the global state database 112 also allows a quick and
immediate migration of the flow without affecting the user’s experience. Furthermore,
regular monitoring of the performance data by the agent 118 and the controller 116 facilitates
timely identification of the weak VNAs 106, thus ensuring smooth and continuous operation
of the cloud.
[0041] Further, in one embodiment, the flow distribution system 110 may facilitate
scaling up and scaling down of the VNAs 106 in the cloud 104. Scaling up of the VNAs 106
may be understood as the process of increasing the number of VNAs 106 present in the cloud
104 in order to either handle an increase or a potential increase in the traffic, i.e., the number
of flows or data packets handled by the cloud 104 or reduce load on an existing overloaded
VNA 106. Scaling down of the VNAs 106 may be understood as the process of the reducing
the number of VNAs 106 present in the cloud 104 in order to reduce the resources utilized by
the cloud in case the load currently handled by the cloud 104 can be still be handled by the
VNAs 106 remaining after removal of one VNA 106. Thus, scaling up or scaling down the
VNAs 106 may facilitate the network appliances managing architecture 108 in efficiently
managing the resource utilization of the cloud 104.
[0042] In one implementation, the controller 116 may initially ascertain total load,
e.g., the total number of data packets or flows handled by the VNAs 106. For instance, the
controller 116 may identify the load individually handled by each of the VNA 106 and add
the load to determine the total load handled by all the VNAs 106 together. The controller 116
15
may subsequently compare the total load with a minimum threshold level and a maximum
threshold level of load that can be handled by the VNAs 106. The minimum threshold level
defines the minimum load that a particular number of VNAs 106 should handle in order to
achieve efficient resource utilization in the cloud 104. The maximum threshold level defines
the maximum load that a particular number of VNAs 106 may handle in order to achieve
efficient flow processing with adequate resource utilization in the cloud 104.
[0043] Based on the comparison, the controller 116 may determine whether either of
the scaling up or scaling down has to be performed. In case the controller 116 determines to
scale down the VNAs 106, the controller 116 may obtain the performance data of all the
VNAs 106 to ascertain the VNA 106 that may be removed after migrating its flows to other
VNAs 106. The controller 116 may then determine the performance status of all the VNAs
106 to identify the VNA 106 having a weakest performance status as the VNA 106 that may
be removed. In one implementation, the performance status of each VNA 106 may be
analyzed based on the number of flows handled by the VNA 106 and the resources, such as
processor and memory space utilized by the VNA 106. The controller 116 may subsequently
send the flow migration request to the classifier 114 for migrating the flows of a weak VNA
106 thus identified. The classifier 114 in turn may migrate the flows of the weak VNA 106 to
other VNAs 106 based on the predetermined mapping policy and then remove the weak VNA
106.
[0044] In case the controller 116 decides to scale up the VNAs 106, the controller 116
may launch a new VNA to which the load of one or more existing flows may be migrated, in
order to balance load on all the VNAs 106. The controller 116 may then obtain the
performance data of all the VNAs 106 to ascertain at least one VNA 106 whose flows may be
migrated to the other VNAs 106. For the purpose, the controller 116 may initially determine
an aggregate load of the cloud 104 by dividing the total load by the number of VNAs 106
present in the cloud 104. The controller 116 may then identify the VNAs 106 handling load
greater than the aggregate load and send the flow migration request to the classifier 114 for
migrating the flows of the VNA 106 thus identified. The classifier 114 in turn may migrate
the flows of the identified VNA 106 to other VNAs 106 based on the predetermined mapping
policy such that all the VNA 106 present in the cloud 104 handle load less than or equal to
the aggregate load of the cloud 104. The flow distribution system 110 may thus efficiently
manage the VNAs 106 and the flows handled by the VNAs 106 in the cloud 104.
16
[0045] Although the performance of scaling up and scaling down the VNAs 106 has
been described with reference to an automatic monitoring of the VNAs 106 by the controller,
it will be understood by a person skilled in the art that scaling up or scaling down may be
performed upon receiving instructions from a service provider of the cloud 104.
[0046] Further, in on implementation, the controller 116 and the classifier 114 may
also include agents similar to the agent 118 in order to monitor the controller 116 and the
classifier 114, respectively. Furthermore, in case the classifier 114 fails during the operation
of the cloud 104, the controller 116 may request the VNAs 106 to provide a list of the flows
handled by them. Based on the lists obtained from each of the VNAs 106, the controller 116
may reconstruct the mapping for each of the VNAs 106 and provide details of the mappings
to a new virtual machine assigned to operate as the classifier 114. The network appliances
managing architecture 108 thus facilitates in avoiding all possible failures of the cloud 104.
[0047] Figure 2 and 3 illustrate a method 200 and a method 300, respectively, for
managing virtual network appliances in a cloud computing networking, according to an
embodiment of the present subject matter. The order in which the method is described is not
intended to be construed as a limitation, and any number of the described method blocks can
be combined in any order to implement the methods 200 and 300 or any alternative methods.
Additionally, individual blocks may be deleted from the methods without departing from the
spirit and scope of the subject matter described herein. Furthermore, the method(s) can be
implemented in any suitable hardware, software, firmware, or combination thereof.
[0048] The method(s) may be described in the general context of computer
executable instructions. Generally, computer executable instructions can include routines,
programs, objects, components, data structures, procedures, modules, functions, etc., that
perform particular functions or implement particular abstract data types. The methods may
also be practiced in a distributed computing environment where functions are performed by
remote processing devices that are linked through a communications network. In a distributed
computing environment, computer executable instructions may be located in both local and
remote computer storage media, including memory storage devices.
[0049] A person skilled in the art will readily recognize that steps of the method(s)
200 and 300 can be performed by programmed computers. Herein, some embodiments are
also intended to cover program storage devices or computer readable medium, for example,
digital data storage media, which are machine or computer readable and encode machine17
executable or computer-executable programs of instructions, where said instructions perform
some or all of the steps of the described method. The program storage devices may be, for
example, digital memories, magnetic storage media, such as a magnetic disks and magnetic
tapes, hard drives, or optically readable digital data storage media. The embodiments are also
intended to cover both communication network and communication devices to perform said
steps of the method(s).
[0050] Figure 2 illustrates the method 200 for flow migration between virtual network
appliances (VNAs) in a cloud computing network, according to an embodiment of the present
subject matter.
[0051] At block 202, performance data for a VNA is obtained. In one
implementation, the performance data for a VNA 106 provided in a cloud computing
network, say, the cloud 104 may be obtained by a controller, say, the controller 116. The
performance data may include values of one or more performance parameters, such as
processor utilization, memory utilization, and number of flows handled by the VNA. In one
embodiment, the performance data may be received as a part of a periodical update from the
VNA. In another embodiment, the performance data may be received upon a request from the
controller.
[0052] At block 204, a determination is made to ascertain whether the VNA has a
weak performance status or not. In one implementation, the performance data may be
analyzed by the controller to determine the performance status of the VNA. If the controller
determines the performance status to be a balanced loaded status, the controller may
determine the VNA as not having the weak performance status, which is the 'No' path from
the block 204, the method moves back to the block 202, where the performance data may be
further received, for example, after a predetermined time period.
[0053] In case at block 204 it is determined that the VNA has the weak performance
status, i.e., the VNA has any one of an under-loaded, overloaded, and failed status, which is
the 'Yes' path from the block 204, a flow migration request for migrating one or more flows
of data packets from is provided to a classifier, say, the classifier 114 at block 206. For
example, on determining the VNA to have a weak performance status the controller may
decide to remove the VNA in order to manage resource utilization in the cloud 104. The
controller may thus send the flow migration request to the classifier asking the classifier to
18
migrate the flows of the VNA to another VNA. Further, in case the performance status of the
VNA is either of the overloaded or the failed status, the controller may launch a new VNA to
which the classifier may migrate one or more of the flows of the VNA.
[0054] At block 208, at least one active VNA is identified for flow migration based
on a predetermined mapping policy. In one implementation, upon receiving the flow
migration request from the controller, the classifier may identify the at least one active VNA,
from among a plurality of VNAs, having the performance status corresponding to one of the
under-loaded and balanced loaded status. In another implementation, the classifier may
identify the new VNA as the active VNA in case the controller indicates the launching of the
new VNA in the flow migration request. Further, the classifier may determine the at least one
active VNA based on the predetermined policies, such as round robin policy.
[0055] At block 210, one or more flows from the VNA are migrated to the at least
one active VNA. Upon identifying the at least one active VNA, the classifier may send flow
mapping instructions to the at least one active VNA indicating the migration of the one or
more flows from the VNA. Upon receiving the flow mapping instructions, the at least one
active VNA may obtain global state of the flows and start processing of the flows. Further, in
case the performance status of the VNA whose flows are migrated was under-loaded, then the
classifier may remove the VNA upon migrating the flow to the at least one active VNA.
[0056] Figure 3 illustrates the method 300 for managing virtual network appliances,
according to an embodiment of the present subject matter
[0057] At block 302, total load handled by a plurality of VNA operating in a cloud
computing network is ascertained. In one implementation, the total load handled by each of
the plurality of VNA may be obtained and a sum of the load handled by the VNAs. VNA
may be ascertained to obtain the total load handled by the plurality of VNA in the cloud
computing network, for example, the cloud 104.
[0058] At block 304, a determination is made to ascertain whether the total load is
less than a minimum threshold level. For instance, the total load is compared with the
minimum threshold level. If the controller 116 determines that the total load is greater than
the minimum threshold level which is the 'No' path from the block 304, a determination is
made at block 306 to ascertain whether the total load is greater than a maximum threshold
level. For instance, the total load is compared with the maximum threshold level. If the
controller 116 determines that the total load is less than the maximum threshold level which
19
is the 'No' path from the block 306, the method moves back to the block 302, where the total
load may be re-ascertained, for example, after a predetermined time period.
[0059] In case at block 306 it is determined that the total load is greater than the
maximum threshold level, which is the 'Yes' path from the block 306, it is determined to
perform scaling up of the plurality of VNA at block 308.
[0060] In case at block 304 it is determined that the total load is less than the
minimum threshold level, which is the 'Yes' path from the block 304, it is determined to
perform scaling down of the plurality of VNA at block 310.
[0061] On determining to perform either of scaling up or scaling down at the block
308 and 310, respectively, the method moves at block 312. At the block 312, at least one
VNA is identified from among the plurality of VNA for flow migration based on the
determination. In one implementation, upon determining to perform scaling up, at least one
VNA may be ascertained that is handling load greater than an aggregate load of the plurality
of VNA and identified as the at least one VNA for flow migration. In another
implementation, upon determining to perform scaling down, at least one VNA may be
ascertained that is having the weakest performance status among the plurality of VNA and
thus identified as the at least one VNA for flow migration, a new VNA may be launched
Alternatively, as discussed in method 200, the controller may decide to migrate flows from
one VNA to another VNA upon determining the VNA to have weak performance status.
[0062] At block 314, a flow migration request for migrating one or more flows of
data packets from the at least one VNA is provided to a classifier. For example, on
determining to perform scaling down, the controller may decide to remove the at least one
VNA in order to manage resource utilization in the cloud 104. The controller may thus send
the flow migration request to the classifier asking the classifier to migrate the flows of the at
least one VNA to another VNA. Further, on determining to perform scaling up, the controller
may decide to reduce load of the at least VNA and launch a new VNA. The controller may
thus send the flow migration request to the classifier asking the classifier to migrate the flows
of the at least one VNA to the new VNA.
[0063] At block 316, the one or more flows from the at least one VNA are migrated
to at least one active VNA. Upon identifying the at least one active VNA, the classifier may
send flow mapping instructions to the at least one active VNA indicating the migration of the
20
one or more flows from the at least one VNA. Upon receiving the flow mapping instructions,
the at least one active VNA may obtain global state of the flows and start processing the
flows. Further, in case of scaling down, the classifier may remove the at least one VNA upon
migrating the flow to the at least one active VNA.
[0064] Although primarily depicted and described in a particular sequence, it should
be appreciated that the steps shown in methods 200 and 300 may be performed in any
suitable sequence. Moreover, the steps identified by one step may also be performed in one or
more other steps in the sequence or common actions of more than one step may be performed
only once. For example, step 306 may be performed before step 304 or steps 304 and 306
may be performed at the same time and an indicator may determine whether the method
continues to step 302, 308 or 310.
[0065] Although embodiments for flow migration between virtual network appliances
in the cloud computing network have been described in a language specific to structural
features or method(s), it is to be understood that the invention is not necessarily limited to the
specific features or method(s) described. Rather, the specific features and methods are
disclosed as embodiments for flow migration between virtual network appliances in the cloud
computing network.
21

I/We claim:
1. A network appliances managing architecture (108) for migrating flow between virtual
network appliances (VNAs) in a cloud computing network (102), the network
appliances managing architecture (108) comprising:
a controller (116) to:
obtain performance data for a VNA (106); and
analyze the performance data to determine whether the VNA (106) has
a weak performance status, where the weak performance status corresponds to
any one of an overloaded, an under-loaded, and a failed status; and
a classifier (114) to:
receive a flow migration request from the controller (116) for
migrating one or more flows of data packets from the VNA (106) based on the
analyzing;
identify an active VNA (106) for flow migration based on a
predetermined mapping policy; and
migrate the one or more flows from the VNA (106) to the at least one
active VNA (106).
2. The network appliances managing architecture (108) as claimed in claim 1, wherein
the active VNA (106) further:
receives flow mapping instructions from the classifier (114) for initiating
handling of the migrated flow of data packets;
obtains global state for the migrated flow from global state database (112); and
initiates handling of the new flow for processing the data packets based on the
global state.
3. The network appliances managing architecture (108) as claimed in claim 2, wherein
the active VNA (106) comprises an agent (118) to:
obtain the global state for the migrated flow from the global state database
(112); and
periodically update local state of the migrated flow to the global state database
(112).
4. The network appliances managing architecture (108) as claimed in claim 1, further
comprising a plurality of VNAs (106), wherein each of the plurality of VNAs (106)
further comprises an agent (118) to:
22
monitor performance parameters of the corresponding VNA (106); and
provide the performance data to the controller (116) based on the monitoring.
5. The network appliances managing architecture (108) as claimed in claim 1, wherein
the controller (116) further launches a new VNA for the performance status of the
VNA (106) corresponding to one of the overloaded and the failed status.
6. The network appliances managing architecture (108) as claimed in claim 1, wherein
the classifier (114) further:
determines at least one VNA (106), from among a plurality of VNA (106),
having the performance status corresponding to one of the under-loaded and a
balanced loaded status as the at least one active VNA (106); and
removes the VNA (106) upon flow migration to the at least one active VNA
(106).
7. A method for flow migration between virtual network appliances (VNAs) in a cloud
computing network, the method comprising:
obtaining performance data for a VNA;
analyzing the performance data to determine whether the VNA has a weak
performance status, where the weak performance status corresponds to any one of an
overloaded, an under-loaded, and a failed status;
providing a flow migration request to a classifier for migrating one or more
flows of data packets from the VNA based on the analyzing;
identifying at least one active VNA for flow migration based on a
predetermined mapping policy; and
migrating the one or more flows from the VNA to the at least one active VNA.
8. The method as claimed in claim 7, wherein the method further comprises launching a
new VNA for the performance status of the VNA corresponding to one of the
overloaded and the failed status.
9. The method as claimed in claim 8, wherein the identifying further comprises
determining the new VNA as the at least one active VNA for flow migration.
10. The method as claimed in claim 7, wherein the identifying, for the performance status
corresponding to one of the overloaded and the failed status, further comprises
determining at least one VNA, from among a plurality of VNAs, having the
performance status corresponding to one of the under-loaded and a balanced loaded
status as the at least one active VNA.
23
11. The method as claimed in claim 7, wherein the method further comprises removing
the VNA upon flow migration to the at least one active VNA for the performance
status corresponding to the under-loaded status.
12. The method as claimed in claim 7, wherein the method further comprises obtaining,
by the at least one active VNA, global state corresponding to the migrated flow from
global state database to initiate processing of the data packets corresponding the
migrated flow.
13. A method for managing virtual network appliances (VNAs) comprising:
ascertaining total load handled by a plurality of VNA operating in a cloud
computing network;
comparing the total load with a minimum threshold level and a maximum
threshold level;
determining whether to perform at least one of a scaling up or a scaling down
of the plurality of VNAs based on the comparing;
identifying at least one VNA from among the plurality of VNA for flow
migration based on the determination;
providing a flow migration request to a classifier for migrating one or more
flows of data packets from the at least one VNA based on the identifying; and
migrating the one or more flows from the at least one VNA to at least one
active VNA based on a predetermined mapping policy.
14. The method as claimed in claim 13, wherein the identifying, for performing scaling
down, further comprises:
obtaining performance data for each of the plurality of VNA, wherein the
performance data indicates value of one or more performance parameters of the
corresponding VNA, and wherein the one or more performance parameters include
processor utilization, memory utilization, and number of flows handled;
analyzing the performance data to determine performance status of each of the
plurality of VNA; and
identifying a VNA having a weakest performance status as the at least one
VNA.
15. A computer-readable medium having embodied thereon a computer program for
executing a method of flow migration between virtual network appliances (VNAs) in
a cloud computing network, the method comprising:
24
receiving performance data for a VNA;
analyzing the performance data to determine whether the VNA has a weak
performance status, where the weak performance status corresponds to any one of an
overloaded, an under-loaded, and a failed status;
providing a flow migration request to a classifier for migrating one or more
flows of data packets from the VNA based on the analyzing;
identifying at least one active VNA for flow migration based on a
predetermined mapping policy; and
migrating the one or more flows from the VNA to the at least one active VNA.
Date 12 April 2013
DAMODAR PANDHARINATH VAIDYA
IN/PA-1431
Agent for the Applicant
To,
The Controller of Patents
The Patent Office at New Delhi
2