Feedback
Table Of Contents
Configuring Dynamic Workload Scaling
Overview of Dynamic Workload Scaling
Configuring Dynamic Workload Scaling
Configuration Guidelines and Operating Considerations
Configuring the IP Address of the Nexus Device
Configuring the Credentials of the Nexus Device
Configuring the VM Controller on the ACE
Configuring the URL of the VM Controller
Configuring the Credentials of the VM Controller
Configuring a VM Probe on the ACE
Configuring the VM Probe Interval
Configuring the VM Controller for the VM Probe
Configuring the Load Type and Bursting Threshold for the VM Probe
Configuring Virtual Machines as Real Servers
Configuring Server Farm Attributes for DWS
Enabling a Server Farm for Dynamic Workload Scaling
Associating the VMs as Real Servers with the Server Farm
Displaying Dynamic Workload Scaling Statistics
Displaying the Nexus Device Statistics
Displaying the VM Controller Statistics
Displaying the DWS State of the VIP
Displaying the DWS State of the Server Farm and the Locality of the Real Servers in the Server Farm
Displaying the Locality of the Real Servers
Displaying the VM Probe Statistics
Displaying the VM Probe Details
Configuring Dynamic Workload Scaling
This chapter describes the dynamic workload scaling (DWS) feature and how to configure it on the ACE appliance. It contains the following major sections:
•
Overview of Dynamic Workload Scaling
•
•
Overview of Dynamic Workload Scaling
DWS integrates the ACE with both the Cisco Nexus 7000 series switches that are running the Cisco Overlay Transport Virtualization (OTV) Data Center Interconnect (DCI) Technology and VMware vCenter Server 4.0 or later. The OTV technology allows the extension of the Layer 2 data center connection into the virtual private cloud (VPC) space. The VMWare technology allows the ACE to retrieve virtual machine load information through the VSphere web service APIs. This combination of hardware and software provides a complete solution for DWS across data centers. When resource demands exceed the capacity of the local data center, this feature enables the dynamic allocation of resources to a remote data center in the VPC. You can configure the ACE to load balance traffic only to the local virtual machines (VMs) or to a combination of local and remote VMs.
Figure 6-1 shows how ANM and the ACE can be integrated with the Cisco Nexus 7000 running OTV software, VMware vCenter (VM controller), and VMs to allow you to create and manage server farms for application delivery that consist of a combination of physical servers and VMs. This network topology illustrates the DWS feature that permits on-demand access to remote resources that you already own, or that you rent or lease from an Internet service provider (ISP) or a cloud service provider.
Figure 6-1 Dynamic Workload Scaling Topology
DWS uses the Cisco Nexus 7000 series switch with Cisco OTV technology device to create a capacity expansion link (Layer 2 link) over an existing IP network between geographically distributed data centers. The Nexus 7000 in the local data center contains an OTV forwarding table that lists the MAC addresses of the Layer 2 extended virtual private network (VPN) and identifies the addresses as either local or remote. The ACE uses an XML query to poll the Nexus 7000 and to obtain the OTV forwarding table information to determine the locality of the VMs (local or remote).
The ACE uses a health-monitoring probe of type VM to poll the VM controller through the vSphere API to obtain the loads of the local VMs. The VM probe is similar to the ACE DNS probe except that the configurable interval between probes has a larger default value (300 seconds). The VM probe maintains a table of the IP addresses of the local VMs and uses persistent SSL connections on port 443 to connect to the VM controller. For details, see the "Configuring a VM Probe on the ACE" section.
The load of the local VMs is based on CPU usage, memory usage, or both. The ACE calculates the average aggregate load of the VMs from the load information returned by the probe. When the average CPU usage or memory usage of the local VMs is greater than or equal to the configured maximum threshold value, the ACE bursts traffic to the remote VMs. When the local VM average CPU and memory loads drop below their configured minimum threshold values, the ACE stops bursting traffic to the remote VMs and continues to load balance traffic locally.
Configuring Dynamic Workload Scaling
This section describes how to configure DWS. It contains the following subsections:
•
•
•
•
•
Prerequisites
DWS requires the following configuration elements:
•
•
•
•
•
•
•
Configuration Guidelines and Operating Considerations
Use the following configuration guidelines when you are configuring and using the DWS feature on the ACE:
•
•
•
•
•
Configuring the Nexus Device
The ACE uses the local Nexus 7000 series switch to determine the locality of the VMs in the local and the remote data centers. You can configure only one Nexus 7000 per ACE. Use the nexus-device command in configuration mode in the Admin context to create the local Nexus device on the ACE.
Note
The syntax of this command is as follows:
nexus-device name
The name argument specifies the identifier of the local Nexus 7000 that the ACE queries for the locality information of the VMs.
For example, to create the Nexus device named NEXUS_1, enter the following command:
host1/Admin(config)# nexus-device NEXUS_1host1/Admin(config-dci)#The ACE enters DCI configuration mode.
To remove the Nexus device and all its attributes from the ACE configuration, enter the following command in configuration mode:
host1/Admin(config)# no nexus-device NEXUS_1This section contains the following topics:
•
•
Configuring the IP Address of the Nexus Device
Use the ip-address command in DCI configuration mode to specify the IP address of one of the management interfaces of either the default VDC on the Nexus 7000 or the VDC on which the OTV configuration is deployed. This address is the one that ACE uses to connect to the Nexus 7000 to discover the locality of the VMs. The syntax of this command is as follows:
ip-address ip_address
The ip_address argument specifies the IP address of the Nexus 7000 management interface in the local data center.
For example, to configure the IP address of the Nexus device located at 192.168.12.15, enter the following command:
host1/Admin(config-dci)# ip-address 192.168.12.15To remove the IP address of the Nexus device from the ACE configuration, enter the following command:
host1/Admin(config-dci)# no ip-address 192.168.12.15Configuring the Credentials of the Nexus Device
Use the credentials command in DCI configuration mode to configure the username and password of the Nexus device on the ACE. The ACE uses these credentials to log in to the Nexus 7000.
Note
The syntax of this command is as follows:
credentials {username} {[encrypted] password}
The keywords and arguments are as follows:
•
•
•
Note
For example, enter the following command:
host1/Admin(config-dci)# credentials admin encrypted mydcipassphraseTo remove the credentials of the Nexus device from the ACE configuration, enter the following command:
host1/Admin(config-dci)# no credentials admin encrypted mydcipassphraseConfiguring the VM Controller on the ACE
The ACE uses the VMware vCenter Server 4.0 or later as the VM controller to obtain the aggregate load data of the VMs in the local data center. You can configure only one VM controller per ACE. Use the vm-controller command in configuration mode in the Admin context to configure the VM controller.
Note
The syntax of this command is as follows:
vm-controller name
The name argument specifies the name of an existing VM controller (vCenter) that the ACE queries for VM load information. Enter an unquoted text string with no spaces and a maximum of 64 alphanumeric characters.
For example, enter the following command:
host1/Admin(config)# vm-controller VCENTER_1host1/Admin(config-vm)#The ACE enters VM configuration mode.
To remove the VM controller and all its attributes from the ACE configuration, enter the following command in config mode:
host1/Admin(config)# no vm-controller VCENTER_1This section contains the following topics:
•
•
Configuring the URL of the VM Controller
Use the url command in VM configuration mode to configure the location of the VM controller in the network. The syntax of this command is as follows:
url string
The string argument specifies the host name or the IP address of the VM controller. Enter an unquoted text string with no spaces and a maximum of 255 alphanumeric characters. You must append "/sdk" at the end of the host name or IP address string.
For example, enter the following command:
host1/Admin(config-vm)# url https://192.168.12.15/sdkTo remove the URL of the VM controller from the ACE configuration, enter the following command:
host1/Admin(config-vm)# no url https://192.168.12.15/sdkConfiguring the Credentials of the VM Controller
Use the credentials command in VM configuration mode to configure the username and password of the VM controller on the ACE. The syntax of this command is as follows:
credentials {username} {[encrypted] password}
The keywords and arguments are as follows:
•
•
•
For example, enter the following command:
host1/Admin(config-vm)# credentials admin encrypted myvmpassphraseTo remove the credentials of the VM probe from the ACE configuration, enter the following command:
host1/Admin(config-vm)# no credentials admin encrypted myvmpassphraseConfiguring a VM Probe on the ACE
The ACE uses a probe of type vm to poll the local VM load information from the VM controller. The ACE calculates the average aggregate load information as a percentage of CPU usage or memory usage to determine when to burst traffic to the remote data center. If the server farm consists of both physical servers and VMs, the ACE considers load information only from the VMs. After you configure the VM probe and its attributes, you associate it with a VM controller and a server farm. Use the probe command in configuration mode to configure a probe of type vm. The syntax of this command is as follows:
probe probe_type probe_name
The arguments are as follows:
•
•
For example, enter the following command:
host1/Admin(config)# probe vm VM_PROBEhost1/Admin(config-probe-vm)#To remove the VM probe and all its attributes from the ACE configuration, enter the following command:
host1/Admin(config)# no probe vmThis section contains the following topics:
•
•
•
Configuring the VM Probe Interval
Use the interval command in probe VM configuration mode in the Admin or user contexts to specify the frequency with which the ACE sends probes to the VM controller. The syntax of this command is:
interval value
The value argument specifies the elapsed time between probes. Enter the time interval in seconds as an integer from 300 to 65535. The default is 300 seconds (5 minutes).
For example, to configure the ACE to send a probe to the VM controller every 420 seconds (7 minutes), enter the following command:
host1/Admin(config-probe-vm)# interval 420To reset VM probe interval to the default value of 300 seconds (5 minutes), enter the following command:
host1/Admin(config-probe-vm)# no intervalConfiguring the VM Controller for the VM Probe
Use the vm-controller command in probe VM configuration mode in the Admin or user contexts to identify the VM controller for the probe. The syntax of this command is as follows:
vm-controller name
The name argument is the identifier of the VM controller that you configured in the "Configuring the VM Controller on the ACE" section. Enter an unquoted text string with no spaces and a maximum of 64 alphanumeric characters.
For example, to configure the VM controller called VCENTER_1, enter the following command:
host1/Admin(config-probe-vm)# vm-controller VCENTER_1To remove the VM controller name from the VM probe configuration, enter the following command:
host1/Admin(config-probe-vm)# no vm-controller VCENTER_1Configuring the Load Type and Bursting Threshold for the VM Probe
Use the load command in probe VM configuration mode in the Admin context or user contexts to specify the interesting load of the local VMs. You can specify CPU usage, memory usage, or both. The syntax of this command is as follows:
load {cpu | mem} burst-threshold {max value min value}
The keywords and arguments are as follows:
•
•
If the average load value returned by the VM controller is greater than or equal to the maximum threshold value, the ACE starts bursting traffic to the remote data center. When the load value returned by the VM controller is less than the minimum threshold value, the ACE stops bursting traffic to the remote data center and load balances traffic to the local VMs. Any active connections to the remote data center are allowed to complete.
For example, to instruct the ACE to start bursting traffic to the remote data center when the local average VM load exceeds 80 percent CPU usage and to stop bursting traffic when the local average CPU usage drops below 50 percent, enter the following command:
host1/Admin(config-probe-vm)# load cpu burst-threshold max 80 min 50You can configure an additional load command under the same VM probe to create an OR statement between the CPU usage and the memory usage of the local VMs. Whichever load type reaches its maximum threshold first will cause the ACE to burst traffic to the remote data center. For example, enter the following commands:
host1/Admin(config-probe-vm)# load cpu burst-threshold max 80 min 50host1/Admin(config-probe-vm)# load mem burst-threshold max 70 min 40In this case, if the average CPU usage reaches 80 percent or the average memory usage reaches 70 percent, the ACE bursts traffic to the remote data center. The ACE does not stop bursting traffic to the remote data center until both the CPU load and the memory load drop below their respective minimum configured values.
To reset the VM probe behavior to the default of checking the average VM CPU usage and memory usage against the maximum and minimum threshold values of 99 percent each, enter the following command:
host1/Admin(config-probe-vm)# no load cpu burst-threshold max 80 min 50host1/Admin(config-probe-vm)# no load mem burst-threshold max 70 min 40Configuring Virtual Machines as Real Servers
To allow the ACE to load balance traffic to the local and remote VMs, configure the VMs as real severs on the ACE. Use the rserver command in configuration mode and then associate the VMs with a server farm. For details about configuring real servers and server farms, see Chapter 2, Configuring Real Servers and Server Farms.
Configuring Server Farm Attributes for DWS
Use the serverfarm command to configure a server farm for the DWS feature. Only server farms of type host are supported by DWS. The syntax of this command is as follows:
serverfarm host name
For example, to configure an existing server farm called DWS_SF1 for DWS, enter the following command:
host1/Admin(config)# serverfarm host DWS_SF1host1/Admin(config-sfarm-host)#This section includes the following topics:
•
•
Enabling a Server Farm for Dynamic Workload Scaling
Use the dws command in server farm host configuration mode in the Admin or a user context to enable the server farm for the DWS feature. The syntax of this command is as follows:
dws {local | burst probe name}
The keywords and arguments are as follows:
•
•
•
•
For example, to configure a server farm for DWS, enter the following command:
host1/Admin(config-sfarm-host)# dws burst probe VM_PROBETo disable DWS on a server farm, enter the following command:
host1/Admin(config-sfarm-host)# no dws burst probe VM_PROBEAssociating the VMs as Real Servers with the Server Farm
To complete the server farm configuration for DWS, you must associate the VMs as real servers in the server farm. Use the rserver command in server farm host configuration mode. For details about configuring server farms and associating real servers with them, see Chapter 2, Configuring Real Servers and Server Farms.
Displaying Dynamic Workload Scaling Statistics
This section describes the show commands that you can use to display DWS statistics. It contains the following subsections:
•
•
•
•
•
•
•
Displaying the Nexus Device Statistics
You can display the Nexus device connection statistics by using the show nexus-device command in Exec mode in the Admin context. The syntax of this command is as follows:
show nexus-device [name] [detail]
The argument and keyword are as follows:
•
•
For example, to display the Nexus device statistics, enter the following command:
host1/Admin# show nexus-device DC1 detailTable 6-1 describes the fields in the show nexus-device command output.
Displaying the VM Controller Statistics
You can display the VM controller connection statistics by using the show vm-controller command in Exec mode in the Admin context. The syntax of this command is as follows:
show vm-controller [name] [detail]
The argument and keyword are as follows:
•
•
For example, to display the VM controller connection statistics, enter the following command:
host1/Admin# show vm-controller VCENTER1 detailTable 6-2 describes the fields in the show vm-controller detail command output.
Displaying the DWS State of the VIP
You can display the DWS state of a VIP by using the show service-policy name command in Exec mode. The syntax of this command is as follows:
show service-policy name
Table 6-3 describes the DWS-related field in the show service-policy name command output.
Table 6-3 DWS-Related Field Descriptions for the show service-policy name Command Output
VIP DWS state
The DWS state of the VIP. Possible states are DWS_ENABLED or DWS_DISABLED.
For details about the other fields in the show service-policy name command output, see Chapter 3, Configuring Traffic Policies for Server Load Balancing.
Displaying the DWS State of the Server Farm and the Locality of the Real Servers in the Server Farm
You can display the DWS state of the server farm and the locality of the real servers in the server farm by using the show serverfarm name command in Exec mode. The syntax of this command is as follows:
show serverfarm name
Table 6-4 describes the DWS-related fields in the show serverfarm name command output.
For details about the other fields in the show serverfarm name command output, see Chapter 2, Configuring Real Servers and Server Farms.
Displaying the Locality of the Real Servers
You can display the locality of the VMs by using the show rserver name command in Exec mode. The syntax of this command is as follows:
show rserver name
Table 6-5 describes the DCI-related fields in the show rserver name command output.
Table 6-5 DCI-Related Field Descriptions for the show rserver name Command Output
Locality
Specifies whether the VM is LOCAL or REMOTE
For details about the other fields in the show rserver name command output, see Chapter 2, Configuring Real Servers and Server Farms.
Displaying the VM Probe Statistics
You can display the VM probe statistics by using the show stats probe type vm command in Exec mode. The syntax of this command is as follows:
show stats probe type vm
Table 6-6 describes the fields in the show stats probe type vm command output.
Displaying the VM Probe Details
You can display the details of the VM probe settings and statistics by using the show probe name command in Exec mode. The syntax of this command is as follows:
show probe name
Table 6-7 describes the fields in the show probe name command output.
