Part of the answer is in your question:  it is a protocol, and it stands for Locator/ID Separation Protocol.

The use of LISP is targeted at deployment of virtual systems, and the movement/management of those systems/resources (such as Virtual Machines) across networks to other places such as other data centers.

LISP really answers the question a customer might ask: how can I elastically and dynamically move, manage, and adjust compute services across multiple data center locations?

The LISP protocol allows for workload mobility across physical locations, and this is not an easy task.  Having the ability to move end systems such as VMs means you have to manage IP addresses to create flexible cloud resources.  This has been a classical mobility problem in IP that has been solved with clever IP address management, or IP Mobility solutions.

Let me spend a second defining the issue with IP.  An IP address has 2 parts: a network part (IPv4) or prefix (IPv6), and a Host part (IPv4) or interface part (IPv6).  We could redefine these as a "location" and an "end point".  A more colloquial way would be to say a street name and a house number.  So when we move the host or interface and connect to a different router it is likely that the network part, or location is going to change.  There are exceptions to this, of course.  For example if I am attached to my home network with a configured IP address of 192.168.1.175/24, I will almost certainly be able to connect to your home network and not change my IP address.   I can also use IP mobility to manage my movement as well.  

What LISP does is it provides a scalable real world solution to this issue for large scale VM movement across networks where we know the location is changing for a given end point.  LISP does this by creating a database (like a DNS that maps names to IP Addresses) that maps IP addresses of end points to IP addresses of locations.  So what we get is:

  • Correlation of end points to locations
  • Messages that map end points (End Point ID) to Attachment Points (these are usually tunnel end points)
  • The ability to manage and build network policies into the protocol responses
  • A system that reacts automatically via listening to changes in the data plane (watching packets) or via network manager control of the database

There are essentially three components to the LISP enabled network:

  • Switches connected to the VMs or systems are the LISP enabled end points
  • Routers in the network act as LISP enabled proxies
  • Servers run the database

 You can get much more detail here: http://lisp.cisco.com/

Comments powered by CComment