Tip:
Highlight text to annotate it
X
802.11n is a big deal. Big deal for the enterprise networking market, not just for wireless LAN.
Now with 11n, Wi-Fi comes off h. The performance that we've come to expect, the predictability
of per-user connectivity we've come to expect on a wide medium is now possible on wireless.
There's a big pipe, so you have enough capacity. You have capacity almost like fast Ethernet
for every user, but at the same time you also have the reliability, predictability combined
with the mobility to allow users to roam anywhere on your network.
Now, Meru Networks has spent the last five to six years in making that architecture robust,
to make the wireless infrastructure as comparable to wired, but as predictable, as reliable
as wired infrastructure. And that's why thousands of our users the world over are confidently
able to migrate to a wireless LAN to support the machine-critical applications that they
were once running on wired medium. Not just that, they can also run some applications
that they never thought possible on wired.
So while we have made the infrastructure architecture very robust, in the last 12 months we have
concentrated a lot more in actually improving the management products -- the management
and operation products for the wireless LAN. Now if you really look at the enterprise networking
market and try to calculate the cost of a wireless infrastructure, what you'll find
is the useful life of an infrastructure is about three to five years, and typically after
that either you replace your wireless LAN, or there's a next Fi that you need to look
at supporting on your infrastructure.
But the cost of wireless LAN infrastructure or wireless LAN is not limited to infrastructure.
While 40 to 45% of the cost is in infrastructure, it's actually the management and operation
of the network that ends up taking 60 to 65% of your budget in three to five years. So
let's talk about the management piece. There are two pieces here. One is reacting. Reacting
to user demands or user complaints. So while you have an infrastructure that supports all
your mission-critical applications, when the users complain about not being able to connect,
you need to do something to help them out.
These are usually the reactive mechanisms, because the user has complained, you go to
the network management system, you try to find out what is the likely cause. What we
did earlier this year was we introduced an E(z)RF 2.0 Network Management platform, which
allows you to go back, rewind, and recreate your RF environment so you can now find out
what the root cause of the problem was for the user. We also created management signatures,
so not only are the users able to take advantage of other large-scale networks that we have
studied very carefully, we have studied the pattern of different network management events
and have tried to correlate it and create a management signature, they're also able
to save a lot of time so that they don't have to go back every time and find out what the
root cause is. The system tells them that here are the likely causes, here is where
the link is broken, this is what you need to go and focus on. So the time of managing
for the reactive portion of the management was cut by half or more.
Many of our customers have large-scale distribution environments. There is no way that they can
actually go to that location where the user is experiencing a problem, and by the time
they go there, let's say a couple of hours, a couple of days later, the RF environment
has changed. So what the system does is captures all that information, stores it for the user,
and now the Network Management can rewind -- literally rewind -- and say okay, what
happened at Thursday at 4PM? It's something I'm able to see on Monday at 9AM and find
out the root cause and help diagnose the problem in the network. Before it affects the users
in that given area, you're able to do something about it. But what we're introducing into
the market is now the proactive diagnostic. Reactive is all good because it helps you
cut down the troubleshooting costs, it helps you help your users, but it's really proactive
diagnostics that really allows you to come up with a network that you're very confident
before the users show up on your network, you'll be able to do something about fixing
the problem.
So in other words, how do you make sure before your users get on your network, you're able
to rule out any of those root causes that can affect the performance of your wireless
network. As I was mentioning earlier, our wireless LANs are as reliable as wired infrastructure,
but then you have to make sure of the stability. You have to be very confident that I'm providing
a pipe so that my users, when they connect their devices, they're able to run all the
applications the network is designed for.
Let's step back and look at some of the industry examples. If you think about a nurse that's
trying to make a call in an operation theater, let's say at 8AM Monday morning, and finds
the call is not going through. The nurse will contact the IT department, the network manager.
The network manager will try to diagnose the problem, and it may be that it's too late
by the time the network manager is able to help that nurse. Now what happened here was
this call could have been life or death for somebody. Either that nurse had to leave that
operation theater to go back to the central nurse station to take care of the problem,
or had to rely on some other means of communication which was not real time. In either way, the
network actually failed to help or serve that nurse who needed it at that very moment.
Now, how do you make sure this kind of problem doesn't affect your users, and you're able
to make sure that the network is going to operate -- is available for the users before
they come on the network? And that is why we're introducing Services Assurance Manager
for proactive diagnostics.
Now let's go to how the product works. Now, in Meru Networks, all access points are deployed
on a single channel. That's why sometimes it's commonly referred to as single-channel
architecture. Basically, when all your access points are on a single channel, the central
controller is able to coordinate the access between different access points, but also
make sure there is no co-channeling difference by design. But it also provides a large simplicity
in the design, but it also provides a very, very good tool -- a very, very robust architecture
for management. Now let's take a look at how Service Assurance Management, which is a proactive
diagnostic mechanism, takes advantage of the architecture and builds on it further to allow
network managers to be able to centrally diagnose the problem before it affects the end users.
Now, if you look on this picture, you have four access points all deployed on the same
channel. All of these access points are serving the clients, which are shown by the laptops
that are connected. Now, these could be any devices. These could be phones, laptops, just
a normal environment with access points on the same channel on the floor and all access
points serving their clients. Now Services Assurance Manager runs on a services appliance,
SA-1000, and that appliance basically has the software for the manager running on it,
which is two parts. One is a server part, one is a client part. Now basically, that
Services Assurance Manager instantiates a client on one of the access points in the
network. So while this access point is serving other clients, it's able to serve as a client
to neighboring access points. Once this virtual client is instantiated on one access point,
the other access points in the system see it as another client. They cannot tell the
difference whether it's another access point advertising itself as a client, or is it another
laptop that entered the network. A neighboring access point would pick up this virtual client,
will send the packets back to the controller, the controller would send it back to the server
part of Services Assurance Manger.
So now what we have done is we have created this end-to-end loop of Services Assurance
Manager instantiating a client on one of the access points that becomes a virtual client
to the neighboring access point. The neighboring access point picks up that packet, sends those
packets back to the controller, and here you're back in infrastructure. That packet then goes
back to the server. Now the server doesn't have to be on the same network. In fact, it
can be anywhere. It's a layer 3 network, over the Internet, the packet makes it back to
the server. The server is able to tell that yes, this end-to-end packet, this end-to-end
loop was completed and all pipes are clear and all pipes are open -- basically, the network
is open for any kind of communication. What it also does is now you're able to point out
the problems.
If for example the neighboring access point was not available, the server would come to
know. If the controller was down, the server would come to know. If anything was broken
in the back end, like if the RADIUS or DHCP server was not available, the server would
come to know. If the VLAN configurations were not done right, the server would come to know.
So not just that, but many other problems that could affect the user's connectivity,
availability of the network, they become known before the users among your network. And this
is what we're introducing in the market, and this is a proactive way of making sure that
your network is available before your users come on the network.
Now let's look at the term services assurance. Why? I don't like that for a product like
this. It's not just proactively diagnosing. It's not just a proactive diagnostic system.
It also allows you to tie in your business-critical applications or the services that you have
between division to wireless LAN networking. If you run a test on a network, say everyday
in the morning at 8AM and typically the users come in at 9, or if you're trying to run this
test before the students come into a classroom to take an online test, what you're able to
do here is actually give that performance assurance that my network is available before
the users actually come on my network. This is the first time in the industry that we're
able to tie in the business requirements to the wireless LAN infrastructure.