This section details the hardware and software requirements for installing the Cisco Modeling Labs server.

The following table lists hardware requirements that are based on the number of virtual nodes used.

The following tables list the required products for the Cisco Unified Computing System (UCS) C220 M3 Rack Server and the Cisco UCS C460 M1 and M2 Rack Servers on which Cisco Modeling Labs Corporate Edition has been tested.

Note	The following list of equipment is for example purposes only; you can deploy the hardware implementation that best suits your requirements.

Cisco Modeling Labs can be set up in a variety of ways to meet the requirements of end users. Prior to setting up the ESXi server for the Cisco Modeling Labs server, we recommend that you create an installation plan that considers the following factors:

• Provide end-user access to the Cisco Modeling Labs server

  The standard way for end users to access the Cisco Modeling Labs server to create topologies is via HTTP-based connectivity. First, end users log in to the Cisco Modeling Labs server through the Cisco Modeling Labs client GUI. After a simulation is started, end users connect to the specific IP address and port number of the node’s management ports. This is done using either the Cisco Modeling Labs client GUI’s Telnet functionality or using a third-party Telnet client.

  As a system administrator, you should determine if end users will access the Cisco Modeling Labs server only when they are on an internal network, such as, a lab network, or if they need access to the server via the Internet. If end users plan to access it remotely, you should request one or more publicly accessible IP addresses that will be applied to the server.

• Provide direct access to the virtual topologies

  After end users create their virtual topologies and launch their simulations, they may connect to the nodes in the topologies in numerous ways. Understanding the access needs is important for determining the configuration and IP address details for the ESXi server and the Cisco Modeling Labs server.

  There are three access strategies to consider:

  • End users bypass the Cisco Modeling Labs client and connect directly to nodes (OOB Management IP access using FLAT)

    You should consider whether end users will require direct access to the nodes in a running network simulation so that they can enable communication from other devices or software because this will impact your IP addressing scheme. With this option, all nodes can be configured on a reserved management network. All management interfaces are connected to a shared management network segment known as FLAT.

    When OOB access is required, the Cisco Modeling Labs server uses a specific configuration that enables a bridge segment on the Ethernet1 port. External devices that attach to the Ethernet1 port, using the correct IP address are then able to communicate directly with the nodes. The simulation continues to be driven by an end user via the Cisco Modeling Labs client GUI communicating with the Cisco Modeling Labs server at its IP address bound to the Ethernet0 port. The settings.ini file includes IP address details for Ethernet 1. These details can be modified based on your deployment strategy.

  • Inband IP access using FLAT

    Consider this option when end users have to connect to one or more nodes in a running simulation to a physical interface for data-plane traffic. In other words, end users need to pass data-plane and control-plane packets from external devices, such as, routers or traffic generators, into the nodes running in a network simulation. This type of connection option will impact your IP addressing scheme. When enabled, end users assign the FLAT network object in the GUI to an interface, effectively connecting that interface on the node to the network segment marked as FLAT. Using a specific configuration, the Cisco Modeling Labs server provides the FLAT network through a bridge segment that connects to the Ethernet1 port.

    External devices attached to the Ethernet1 port with the correct IP addresses are able to pass packets into the destination nodes. A distinct OOB management network is still maintained, but will not be accessible at the same time as the in-band data-plane access. The simulation continues to be driven by the user via the Cisco Modeling Labs client GUI communicating with the Cisco Modeling Labs server at its IP address bound to the relevant management port. The settings.ini file includes IP address details for Ethernet 1. These details can be modified based on your deployment strategy.

    When using FLAT, the node can ping, connect via Telnet, trace a route directly to an external device and vice versa, as long as the target device is on the same subnet, or, if the node has the correct gateway address, and the necessary routing entries, and the subnet that the node has an address on, is a reachable address space from the target device. In other words, the target device needs to know how to communicate back to the node.

  • Inband access using SNAT

    Alternatively, the Static NAT (SNAT) approach provides similar functionality to the FLAT approach, the key difference being that an Openstack-provided and controlled function will translate inbound and outbound packet IP addresses. An internal address and an external address are assigned, for example, 10.11.12.1, which is assigned as the internal address, is mapped to 172.16.2.51 externally. Traffic sent to 172.16.2.51 will be translated to the correct internal address and presented to the node.

    From a UI perspective, the internal and external addresses being used by each node appear in the simulation perspective. The settings.ini file includes IP addressing details for Ethernet 2, which is the port predefined for SNAT. The addressing details can be modified based on your deployment strategy.

• Determine your IP addressing plan

  The following are the key points to note when determining your IP addressing plan for the ESXi server and the Cisco Modeling Labs server:

  • If end users will be accessing the Cisco Modeling Labs server via the Internet, you will require a publicly accessible address for the server or a router that supports NAT.

  • Related to FLAT or SNAT access, an IP address is required for each node being run on the Cisco Modeling Labs server.

    • If the FLAT access method is to be used, then an associated subnet range, sufficient for the number of virtual network devices (Cisco and non-Cisco devices) needs to be allocated. An address range is preconfigured in the settings.ini file. However, it can be modified as needed.

    • If the SNAT access method is to be used, then an associated subnet range, sufficient for the number of virtual network devices (Cisco and non-Cisco devices) needs to be allocated. An address range is preconfigured in the settings.ini file. However, it can be modified as needed.

    You can choose to offer one or the other or both, but in each case, a subnet address range must be provided in order to access the nodes.

  • If you are setting up FLAT or SNAT or both to enable the external devices to connect to the virtual topologies via the Internet, you will need publically accessible IP addresses to be allocated for FLAT and SNAT access methods. 

• Determine if you need to use VLANs in your configurations

  At a minimum, you will require to define VLANs for the management, FLAT, and SNAT networks. You may require more depending on how you plan to segment the network traffic.

• Settings.ini File

  The settings.ini file provides configuration values, such as the IP address ranges to use for FLAT and SNAT nodes, during the initial setup of the Cisco Modeling Labs server. You should follow the installation instructions to set selected parameters during the installation process.

  Caution

  Attempting to change the settings within the settings.ini file after the installation is complete can have adverse effects and leave the server in a non-recoverable state, requiring a reinstallation of the entire OVA/ISO file.

  The following table indicates those settings in the settings.ini file that can be changed once, multiple times, or not at all.

This section details the default port numbers that are provided in Cisco Modeling Labs.

Note

These default port numbers are required for communication between the Cisco Modeling Labs server and the Cisco Modeling Labs client. Therefore, firewalls between the two nodes must be configured to permit these ports to communicate. These values can be updated as required by the system administrator for your Cisco Modeling Labs server installation.