OpenStack Network Concepts

  • by Ying Tang May 25, 2016
  • Tags:

    • OpenStack

    • concepts

This document offers a basic introduction to some common networking concepts. You’ll encounter these concepts when working with your IBM Bluemix Private Cloud.

Terminology Used in This Document

Concept Description
L2 Layer-2 of the ISO 7-layer Internet Architecture, where packets are routed.
VLAN Virtual Local Area Network, essentially a subnet.
IP Internet Protocol, typically used as an abbreviation for an IP Address, which specifices the destination of a packet on the network.
IPv4 Internet Protocol, version 4, an Internet address of the form xxx.xxx.xxx, now usually with the addition of /xx at the end to denote the subnet size using a protocol called CIDR.
CIDR Classless Inter-Domain Routing or supernetting, which is a way to extend the useful life of IPv4 addresses, which can be scarce in a global pool, and therefore usually must be purchased. Here is a link to more information about CIDR: http://searchnetworking.techtarget.com/definition/CIDR. Here is a link to a CIDR subnet size calculator, which is handy for knowing how many IPs you’ll get out of a certain range: http://www.subnet-calculator.com/cidr.php (remember that some of those IPs will be pre-allocated and not available for your cloud).
IPv6 Internet Protocol, version 6, a newer Internet addressing protocol that provides plenty of network addresses but can add complexity to networking architectures.
NIC Network Interface Card, a piece of hardware that attaches servers to a network. Also, Virtual NICs, essentially a piece of software that works like a NIC, are common in cloud usage, and the term “network interface” usually refers to a virtual NIC unless context dictates that it refers to hardware specifically.
MAC The individual identifier, something like a hardware serial number, that uniquely identifies every host on a network.

Basic Networking Concepts

Concept Description
Network An isolated Layer-2 (L2) segment, similar to a VLAN in regular networking. A network may have one or more subnets.
Subnet A range of IPv4 addresses.
Port A connection point for attaching a single device, such as the NIC of a virtual server, to a virtual network. The term “port” also describes the associated network configuration, such as the MAC and IP addresses to be used on that port.
Fixed IP address IP address that is assigned to an instance on boot, such as a network IP address for an instance that needs to reach the public Internet.
Floating IP address IP address that is assigned to an instance on demand, from a pool, often of the IPv6 form.
Quota A quota limits the number of available resources, such as Compute or Disk allocations, for your cloud.
Security groups A set of network filter rules, specific to your project, which are applied to an instance’s network interactions by specifying which IPs can or cannot send packets to another IP or port.
Security group rule A list of protocol, port, and IP address specifications that establish how traffic is allowed flow to, or from, any specific instance on the network; these are the specifics of a security group.

Cloud Network Types

Generally, your cloud will have one or more each of internal and external networks. You can think of this idea as if the instances in your cloud can “talk” to each other on the internal networks, and they “talk” to sources outside your private cloud on the external networks.

Various terms can be used to describe internal and external networks. For example, internal networks sometimes are called “tenant networks.” External networks that are used to communicate with the public Internet sometimes are referred to as “provider networks.”

Internal Networks

An internal network “talks” within your cloud; it provides connectivity among the all of the virtual network interfaces (NICs) attached to it. Each internal network interface is associated with a virtual machine (VM), and a single VM can have multiple network interfaces associated with it, which means it can “talk” on many networks.

You can enable communication among a set of VMs by connecting them to the same internal network. Internal networks are associated with a single project (tenant), and they are visible to and accessible to that project only.

External Networks

External networks can “talk” within and outside your cloud. They may include the public Internet, your cloud’s external network, and the IBM Cloud Data Center network, which contains the IBM Cloud Data Center router.

  • Your cloud’s “public network” or “external network” provides access to and from the public Internet.
  • The SoftLayer private network is intended to provide access to other services in SoftLayer, including other clouds you may wish to communicate with. (See the Release Notes for more information about access to the SoftLayer private network.
  • The SoftLayer router network connects to the SoftLayer Backend Core Router (BCR), which lets you reach resources outside your own private cloud environment, resources which might reside in a data center at your site (Blue Box Local) or in another private cloud environment owned by you, in a SoftLayer data center or elsewhere (Blue Box Dedicated or your own legacy cloud).
  • Public and SoftLayer private networks are physical networks.

Connectivity between Virtual Machines on Different Private Subnets?

Let’s say you have two private networks. You’ve created virtual machines on each network subnet, and you’ve added the necessary security groups so they can accept incoming traffic. How would you achieve connectivity between VMs on different private subnets? Let’s say you have one network named operations and one network called internal. You can achieve communication between them in several ways:

  • Connect your operations network to the default router. This connection enables traffic to flow between the default internal network and the operations network. Now you can deploy an instance to the internal network, assign a public floating IP address to it, and use this instance to connect to the target instance.
  • Or you could use a similar approach and create a new internal router between the internal network and operations network. Use an instance with a public IP on the internal network to give you access to your VM on the operations network. By creating an internal router, you can ensure that VMs on the operations network will not have any external connectivity to the internet.
  • Another possibility is to create a multi-homed instance. Create an instance that is connected to the internal and operations networks. Assign a public floating IP to it. Use this VM to provide access to instances on your operations network through a second, private interface.

What Kind of Security Precautions Should be Taken on Machines with Public Floating IPs?

Here are some good practices to get you started:

  • Configure sshd to permit key authentication only, and disable root ssh access.
  • Do not use standard or default system user names for SSH (ubuntu, deploy and similar).
  • Use a system service like fail2ban or denyhosts to block potential brute force attacks. Both tools are already packaged in most popular Linux distributions.
  • Use security group rules to limit the SSH access to a limited number of IP addresses.
  • Update the system running on your public instance on a regular basis.
  • Do not run any unnecessary services on your public instance (such as http, ftp, smb …)