How are Ethernet packets passed between local and outside networks This task is typically performed by a device called a router, that is also often referred to as a network gateway (in this tutorial the two terms will be used interchangeably). For each Ethernet adapter connected to a computer, the user often specifies an IP address, subnet mask, and default gateway. The default gateway is the address of the router that should be used when packets must be sent outside of the local network.
Most OSs are configured by default to obtain TCP/IP settings (IP address, subnet mask, and default gateway) automatically using a Dynamic Host Configuration Protocol (DHCP) server. If no DHCP server is found, then it is common practice for OSs to assign an IP address in the 169.254.x.x range, which is referred to as a link-local IP address.
When a destination IP address does not fall within the subnet of any NIC, then a default gateway is used to relay packets to the outside network. If multiple default gateways exist, then packets may be routed to the wrong outside network, causing them to be undeliverable.
In most cases, you should be able to address all computers attached to a network interface on the same private subnet, and leave the default gateway blank for that interface. Typically only one NIC with a gateway to the internet or corporate network should have a default gateway address specified.
In other cases where gateways are attached to subnets that connect with different network interfaces, you can either manually specify network routes (this is an advanced practice) or leave the default gateway blank for all interfaces.
While it is more efficient to specify a default gateway on one interface (this avoids the need for an ARP on many packets transmitted to outside networks), leaving all default gateway entries blank can help avoid problems in networks with multiple gateways. However, this strategy will only work if network gateways support proxy ARP.
In this scenario, a host PC is connected to both a corporate network (via NIC #1) which allows internet access as well as a private network (via NIC #2) with several LabVIEW Real-Time targets. The corporate network is setup to assign a DHCP address in the range of 10.0.x.x with subnet mask 255.255.0.0 to NIC #1. The default gateway (router) address is 10.0.0.1.
It is important to leave the default gateway blank on the NIC #2 settings for the host machine. In this way, only the NIC #1 default gateway will be used, which is what we want in order to access the internet from the host. In addition, the default gateway can be left blank on the LabVIEW Real-Time devices, as they should only be communicating within the local subnet (there are no gateways connected).
Assume that the physical connections are used to enable communication to the internet from either OS via a corporate network. The virtual interface will be used only for communication between LabVIEW Real-Time and Windows XP. The IP address of both physical adapters connected to the corporate network will be dictated via DHCP, and will be in the range 10.0.0.x with subnet mask 255.255.255.0. The gateway address is set to 10.0.0.1.
Because we want both Windows XP and LabVIEW Real-Time to access the internet via the physical NICs, they should be the only adapters that have a default gateway set. Therefore, we will leave the default gateway entry for both Virtual Ethernet NICs empty. Note that each OS ultimately has only one default gateway specified.
The Windows physical adapter will still be assigned a DHCP address in the range 10.0.0.x with subnet mask 255.255.255.0. The Virtual Ethernet adapters on Windows XP and LabVIEW Real-Time will once again use static IP addresses of 192.168.0.1 and 192.168.0.2 respectively, with subnet masks of 255.255.255.0 and no default gateway.
To ensure that the physical NIC used with LabVIEW Real-Time is on a different subnet than the Virtual Ethernet NIC, we can use an IP address of the form 192.168.1.x with subnet mask 255.255.255.0. Therefore, the remote PC adapter and LabVIEW Real-Time adapter connected together can use static IP addresses of 192.168.1.1 and 192.168.1.2 respectively with subnet masks of 255.255.255.0. Once again, no default gateway is needed since neither adapter needs to access an outside network (no gateways are present on this subnet).
One more advanced scenario involves configuring a computer (running only one OS this time) with two NICs that communicate with two local networks. Each network features a gateway that can be used to relay packets to additional outside networks.
Imagine that local network A uses IP addresses of the form 10.0.0.x with subnet mask 255.255.255.0, and local network B uses IP addresses of the form 10.0.1.x with subnet mask 255.255.255.0. Local network A is connected via a gateway to outside network C, which uses IP addresses of the form 192.168.0.x with subnet mask 255.255.255.0. Likewise, local network B is connected to outside network D via a gateway, which uses IP addresses of the form 192.168.1.x with subnet mask 255.255.255.0.
The goal is for our multi-NIC PC to be able to access any of the networks. We can connect one NIC to local network A and assign the IP address 10.0.0.5 with subnet mask 255.255.255.0, and we can connect the second NIC to local network B and assign the IP address 10.0.1.5 with subnet mask 255.255.255.0. If any default gateway were assigned, then packets destined for the outside networks may be sent through the wrong gateway, and so this is undesirable.
To solve this problem, we can leave the default gateway entries blank for the two NICs and configure packet routing more manually in the OS. While all OSs are configured differently, most enable the manual addition of routes by the user, and these routes can be configured to persist across reboots of the system. Specifically, we need to add a route with IP address 192.168.0.x and subnet mask 255.255.255.0 to use the IP address of the gateway between networks A and C (e.g. 10.0.0.1). The same needs to be done with IP address 192.168.1.x and subnet mask 255.255.255.0 for the gateway between networks B and D (e.g. address 10.0.1.1).
Additional routes may need to be added as the number of distinct subnets increases. In practice, most networks are set up to avoid the need for this complex configuration by ensuring that each computer is connected to only one gateway.
The Ethernet over InfiniBand driver is available in current releases of the OracleSolaris 11 image. This driver supports the Data Link Provider Interface over allInfiniBand ports of an Oracle Solaris 11 host connected to the gateway. Thedriver uses the IBA unreliable datagram mode to enable initialization, gateway handshake, heartbeat management,frame transmit and receive functions, multicast support, and statistical reporting. The driver isdelivered through an IPS-based package with the file name of ethernet-over-ib.
Administrators create IP interfaces on top of data links. A data link representsa link object in the second layer of the OSI model. The respectivephysical link is directly associated with a device (physical or virtual) and deviceinstance name. The device instance name is comprised of the device driver nameaugmented with the instance number, which has a value of 0 to ninstances of network devices (physical or virtual) using that driver. For each virtual deviceand interface created on the gateway, there is a corresponding data link namedeoibX created on the Oracle Solaris 11 host.
Associating the InfiniBand port of an HCA to a gateway Ethernet port, andassigning one or more MAC addresses to the pair creates a virtual IOadapter (VIOA). The createvnic command of the gateway is used to fulfill thisobjective. Oracle Solaris discovers the VIOA, binds an eoibX datalink instance to the VIOA,and then manages the VIOA as if it were a physical network interfacecard (NIC). Like a physical NIC, the VIOA represents the access path toan Ethernet port. Because more than one MAC address can be assigned toa VIOA, the VNIC management commands of the gateway are actually managing theMAC addresses within the gateway itself.
DNE drivers are not supported with some of the third-party applications even when not in use. For those deployments, admins can use this installation type. As the DNE driver is not installed, only the WFP driver is used for tunneling.
Like many Macintosh users, I've wanted to add a second Ethernet to an iBook (iMac, or Mac Mini) to use as a server (or as a replacement for a dead Ethernet built-in). Although USB to Ethernet adaptors are cheap and plentiful, Mac OS X drivers are not. After scouring the web and talking to other Mac developers, I finally found a workable combination. The purpose of this note is to document what works since I haven't seen it widely published anywhere else. I welcome your feedback to help keep this page up-to-date.
The first driver that worked for me was a Pegasus driver for Mac OS X 10.3 (Panther) developed by Daniel Sumorok. Since then I've been collaborating with Daniel to help make more drivers available. We have ported his original Pegasus driver to work on Mac OS X 10.4 (Tiger or later), and developed a USB 2.0 AX8817x driver. All are open source released under a GPL license.
AX8817x devices are fully backward compatible with USB 1.1 host ports and are reasonably priced, so are probably the best option at this time. Pegasus based devices seem to be getting harder to find. Some USB 1.1 devices use the Realtek 8150 or Davicom DM9601 which is not supported by these drivers.
While other USB-To-Ethernet drivers are reported to be buggy, I haven't encountered any stability problems to date. The adaptor turns off when the computer goes to sleep and comes back on when the computer awakes. It does not support \"Wake On LAN\" at this time.
This USB-To-Ethernet Adaptor combination could be an attractive solution for a Mac Mini, or old iBook used as an Internet gat
0 answers0 replies
P.O. Box 763, 311 S Baughman Road Taylorville, IL 62568