Wi-Fi LANs just keep growing and growing.
And, that’s great, unless you have to manage dozens or hundreds of them one
by one. Then, maybe it’s not so great. Lightweight
Access Point Protocol (LWAPP) an Internet
Engineering Task Force (IETF) draft standard, may provide a remedy for this
management headache.
LWAPP offers an open, vendor-neutral standard to the problem. This proposed
standard is primarily the creation of Airespace, a Wi-Fi network infrastructure
management company (read: WLAN switch vendor), which owns LWAPP related patents.
Airespace is joined by NTT DoCoMo,
the Japanese cell phone giant, and Legra Systems,
another Wi-Fi switch company, in championing LWAPP.
Sounds great? Well, maybe it will be, maybe it won’t. Not only is LWAPP an
open standard at a very early stage of its development, it has some large powerful
business enemies. You see, the 800-pound gorilla of networking, Cisco
has a lot invested in their own proprietary approach to Wi-Fi management, the
Structured Wireless-Aware Network Framework, which powers up Cisco’s CiscoWorks Wireless
LAN Solution Engine 2.0.
This style of management is called ‘Fat AP’ or ‘Peer-to-Peer’ architecture
and this kind of technology is also supported by Proxim. It works well, but
it is expensive and requires access points that can work in a vendor’s particular
infrastructure.
Of course, Airespace, in particular (with its LWAPP-related patents) would
benefit if LWAPP makes it from IETF draft to standard. What we have is a classic
case of a vendor taking their own proprietary policy and trying to make it an
open standard. But since other vendors, especially Cisco, already has a lot
invested in their proprietary standards, LWAPP faces strong opposition. So,
what might appear to some to be a battle between different technology approaches
is really a battle between businesses.
The Ideas of LWAPP
LWAPP is meant to be a network protocol for access points that also provides
for centralized management. The idea for LWAPP started with the observation
that access points work as access servers with IP addresses. This means that
although access points are usually treated as dumb Layer 2 (data link) devices,
they could also work as Layer 3 (network level) devices. For more on layers
see The 7 Layers of
the OSI Model.
LWAPP is meant to be the open, standard protocol for access point management.
In turn, this would be used as the foundation for network management programs
that could be controlled from a switch or router console.
Once deployed, LWAPP’s first goal will be to reduce the filter and policy processing
needed in an access point. That work will be centralized and any changes will
be broadcast to the access points. Then, LWAPP designers will also use this
same centralized management architecture to deal with traffic management, authentication,
encryption, and policy enforcement. Finally, LWAPP will provide a generic encapsulation
and transport mechanism so one vendor’s LWAPP console can work with multiple
vendor’s LWAPP-enabled access points.
At the same time, as management is centralized, LWAPP-equipped access points
would have more memory and processor power so that they could run system access
policies or manage traffic without needing to call-in to a centralized server
for constant instructions.
Does it work? Airespace is already using LWAPP in their commercial AireWave
Director Software. In this application, LWAPP is the underpinning for manual
and automatic access point configuration and management. As for the actual network
management, existing standards like Simple Network Management Protocol (SNMP)
Will LWAPP fly? Sure, the technology is being deployed. The real question is
will it become a real IETF standard. With the opposition of Cisco, that will
not be easy. The draft standard expires in mid-December 2003. By then, we should
know one way or the other.
For now, if you want to experiment with it, you need to invest in Airespace
software and equipment. For the rest of us, we’ll need to wait and see if LWAPP
becomes just another proprietary network management technology or a cure-all
for our network management ailments.
