A guide to application performance management
A comprehensive collection of articles, videos and more, hand-picked by our editors
Enterprise wireless LANs (WLANs) are now used by 80% of corporations -- and most can't afford network latency or slow performance. Learn from industry expert Lisa Phifer how to deploy a high speed wireless network. She describes the network hardware and tools needed to build a high performance WLAN, and she gives strategic planning advice on how to avoid deployment pitfalls. In this screencast, learn how emerging technology improvements, such as 802.11n, can be used to boost network performance so that your business doesn't suffer from a failing WLAN.
Read the full transcript from this video below:
Deploying high performance enterprise wireless LANs, screencast video
Amy Kucharik: Hello and welcome to SearchNetworking.com's webcast on Building High Performance Wireless LANS. My name is Amy Kucharik and I will be the moderator for the duration of this webcast. Joining me today is guest speaker Lisa Phifer. Lisa Phifer owns Core Competence, Inc. a consulting firm, specializing in network security and management technology. She has been involved in the design, implementation and evaluation of network, security, and management products for over 20 years.
At Core Competence, Lisa has advised companies large and small regarding security needs, product assessment and the use of emerging technologies and best practices. Before joining Core Competence, Lisa was a member of the technical staff at Bell Communications Research where she won a President's award for her work on ATM network management. Lisa teaches about wireless LANS, mobile security, and VPNs at many industry conferences and webinars. She has written extensively about network infrastructure and security technologies for numerous publications. Thank you for joining us Lisa. Can you talk to us about what you are going to cover today?
Lisa Phifer: Thanks Amy. In today's webcast I will be talking about how emerging technology improvements like 802.11n can be used to boost performance of enterprise wireless LANS. We will be looking at some of the factors that have to be considered to ensure coverage, capacity and scalability as well as some of the tools that can help you make the most of your investment in next generation [site] throughput generation, wireless LAN equipment.
Amy Kucharik: Whenever you are ready Lisa, you can begin.
Lisa Phifer: Thank you, Amy. Wi-Fi wireless LANs have been successful. While leveraging their early success in the home, wireless networks can be found at 4 out 5 businesses today. According to last year state of the market report about 76% of businesses have deployed Wi-Fi in common areas like meeting rooms and about 59% use Wi-Fi to cover cubicles and office areas. In addition, the number of Wi-Fi client devices have skyrocketed. It's nearly impossible to buy a notebook without Wi-Fi today. The same is expected to happen in mobile phones by the end of this decade.
This graph from Gartner shows, this widely deployed 802.11 bt network and clients actually represent just the tip of the iceberg. Ratification of the 802.11n standard, expected by the end of this year, will trigger even broader wireless deployment. Where some businesses use "BT" today in conference rooms they will probably use "N" to blanket entire floors, buildings and campuses. Client devices will also shift to shipping with "N" by the end of the decade. In fact, this shift has already begun. Over 200 vendors have already released products that comply with the Draft 2.0 of the "N" standard as certified by the Wi-Fi alliance.
As technology matures, businesses will expect more from wireless networks. Companies that deployed "BT" in common areas over the last few years were basically just dipping their toes in the water, learning about the business potential of wireless. Most had fairly low expectations. The top applications now used are internet access and email which require only beset effort data delivery. Use of populations are often relatively modest and users do know they have to be within designated coverage areas in order to use wireless. However, as companies invest in 802.11 upgrades and much broader rollouts they intend to use larger more robust networks to support a wider variety of applications, devices, and users.
According to the state of the market survey cooperative applications like video conferencing and teleconferencing are the number one reason why businesses plan to deploy 802.11n. Those behind are companies expect their new 802.11n network to provide additional capacity for applications like Voice Over IP. Other top drivers prompting end deployment include providing an alternative to wire network access, aggregation of wireless traffic, and support for video streaming applications.
About 23% of those surveyed actually expect 802.11n to reduce the number of access points required. In other words they hope to shrink their ongoing investment in wireless infrastructure. In fact, that's probably a pretty tall order to expect to accomplish network to do more at a lower per user cost. Clearly businesses have high hopes for 802.11n. These expectations are being fueled by what people read about the standard and early products. A maximum data rate defined by 802.11n is roughly 12 times higher than the maximum rate reported by standard HG products. By using MIMO technology and advanced system processing, 802.11n products naturally reach farther. Most vendors claim at least twice the range of "G".
Early pre-end products have been successful in the home, not just because they can extend coverage but also because they can adjust themselves to the RS phenomenon that makes coverage body unreliable. If this network that advantage might not be as important as an increase in total available capacity. These attributes also sound promising, and they are, but 802.11n will not come without some blood, sweat, and tears.
For starters, "n" is going to be a full list upgrade. Companies must invest in new access points and adapters and as we shall see, wired network capacity. Because of the way the 802.11 works, individual client performance will vary depending on where you are standing and what you are holding. If you are upgrading an existing APB network, protection mechanisms will significantly impact and benefit.
Finally the complexity of "n" will raise the bar for installers and administrators requiring more sophisticated tools to cut those tasks down to size. Such challenges are to be expected from any major upgrade. The key is to carefully plan your upgrade to avoid any surprises and too get the most from your investment in the 802.11n.
Every 802.11n device can receive or transmit simultaneously through multiple antennas. Multi input and output antennas are MIMO access point configurations range from 2x2 to 4x4. For example, a 3x3 access point can transmit and receive simultaneously through three antennas. However, it us up to you to decide how to make best use of those antennas by configuring an advanced signal processing options. All end devices are required to support [spatial] multiplexing. Splitting a data stream into pieces and then sending those pieces over different spatial paths through the air then recombining the pieces that are received. This feature increases capacity and data rate.
For example, data sent over two streams can be delivered twice as fast. Many end devices also support an option called space time block coding. In this case, the data that you send is actually sent redundantly over different spatial paths. That gives the receiver better chance of recovering when there is noise or interference that causes errors. If you are using two of three antennas for multiplexing, you have just one leftover for redundancy. Only a handful of "N" access points now offer transmit team format. That's where the sender focuses output and energy in a particular direction based on past performance appears to have better quality delivery to a particular receiver. In other words being poor means using the properties that result in multitask to your advantage.
These physical air improvements are further augmented in the standard by max air improvements. That made the 802.11 protocol the end protocol more air efficient. The 802.11n access point can acknowledge many frames at once, reducing per ax overhead. 802.11n access point can also bunch many smaller frames together into one larger frame and that can reduce the cost of those header bites and interframe spacing.
Finally, 802.11n end devices can transmit over ordinary 20 MHz channels or double wide 40 MHz channels in either 2.4 or 5 GHz spans. We will discuss later, but the bottom line is that all these 802.11n options combined to determine your data rate. There are 77 possible permutations resulting in rates that will range up to 600 mbps when every single end feature is used to its maximum advantage.
Look through some advertising. The biggest misconception about 802.11n high throughput is the meaning of high throughput. To obtain that maximum data rate of 600 mbps, you actually need a pair of 802.11n devices with 4x4 MIMO antennas configured for 64 QAM modulation in both directions using a 40 mg channel and a short 400 second guard interval and no protection to coexist with the ABG. That might sound like a mouthful, but basically it is a lot of options turned on. That maximum configuration might be accomplished its more likely to be used in a business network, at least not in the next year or so. It's much more realistic to think about 300 mbps is the maximum data rate for 2 x 2 MIMO as your theoretical upper bound to start with.
Of course if you deploy ABG wireless you know there is a big difference between data rates as a theoretical number that doesn't count for overhead and idle time and applications performance. Since this chart shows the 802.11b devices at a data rate of 11 mbps typically offer 3 or 4 mb of application throughput. ANG devices that operate at data rates up to 54 mbps max out at about half that number for application throughput. It stands to reason that the same differential is going to hold for 802.11n and in this graph [inaudible 11:27] actually measured 125 mbps for 2 x 2 MIMO access points.
As with all performance metrics your own mileage will vary but if you have the 802.11 end deployment thinking that you are going to 600 or even 300 mb application throughput then you are gong to be disappointed. On the other hand, it is worth pointing out that 125 mbps of throughput is a big number. That is 5 to 6 tines throughput you got with AG at greater distances. It is far more than an application like voice over IP needs anyway. More likely what you should be thinking about is capacity. Higher end data rates means that each message takes less air time to transmit. That means that end access point can deliver more total capacity to be shared amongst all your clients and all your applications.
My point is that it is very important to establish realistic expectations. If you understand what the 802.11 is capable of delivering and the conditions and features that are required to do that, then you stand a better chance choosing the right performance platform. To do this, I recommend developing a concrete plan to build your new high performance WLAN. That plan should be driven by your business objective. If you know what you are trying to achieve, you can develop a phased deployment plan to get there.
To implement your plan you are going to need to identify and satisfy requirements like coexistence in a wire side upgrade. As you move into network design you are going to need to know how many access points to buy, where to put them, and how to configure them in order to achieve your coverage capacity and throughput objectives. Assuming you wish to support multi applications combining data with video then you are going to need to prioritize those wireless streams.
Then finally, you should build management and monitoring tools into your plan from the stat and into your budget so that you will be able to administer your network efficiency and effectively. The rest of this presentation will be looking at each of these steps in further details. I think that eventually every new Wi-Fi product your purchase will support 802.11n. The real question is not whether to deploy N but why, how, and when.
Some companies are investing in draft 2.0 end products already while most are waiting for standard ratification, but even then some will wait for end clients to be more widely deployed. Only you know which is right for you. It isn't easy, but you should start by quantifying your business needs. Identify the specific applications your wireless network will support. For each application specify performance metrics like throughput, latency, jitter, and maximum roam time. Map those applications onto user groups. For example, how many data or voice users will be required to support on each floor and exactly where will they need coverage? As you do that, think about those total users and the number of users active at any one time. Over subscription ratios is 8 to 1 or even higher are pretty common.
Also consider how much redundancy you want to build in your network. For example, should each phone be able to reach two access points and what minimum data rate? Finally, inventory the client devices that you will need to support. For example, how many [Centrino] adapters and how many spectra link and which protocol that is APG do they use today?
Next, use these needs to decide where, how, and when to deploy 802.11n. I listed just a few recommendations here. For example, if your client access needs are already being met, then perhaps N can still help you but what it can help you to do is aggregate that wireless traffic over a back-haul mesh. If your existing users are satisfied but you have business units demanding capacity for new high span applications like video conferencing then consider toying a brand new N network but for just those applications.
On the other hand, if existing users need to reach both old and new applications simultaneously then you are going to have to bite the bullet and deploy bg and n simultaneously on your next generation access point. These are just a few of the strategies for the potential of N.
Next, let's dig a little deeper in this client device performance. Remember that n performance can vary for each individual type of client. Even legacy clients will experience some improvement when they associate n access points but only those new n clients will actually be able to obtain data rates above 54 mgbs, or use those 40 mb channels. Most of your clients use bg in the 2.4 gb hz span today. It is important that you know whether you must support legacy in the 5 GHz span as well. As course your client landscape will evolve over time. Many companies refresh their notebooks every 2 or 3 years and I suggest that all will support N by the year end 2010.
However, actually some users will support high support sooner. How will that change your migration plan? On the flip side most companies are going to be forced to deal with the lagers particularly small devices with embedded Wi-Fi clients. That means you probably won't be able to go dark on Bg until those devices can be retired or replaced. Fortunately, 802.11n standards provide for peaceful coexistence with legacy abg devices. In the near term many new n access points will operate in what's called high throughput mix mode, a mode that uses an extra preamble and protection mechanism so that abg devices can actually share the air peacefully with N devices. Because this mechanisms have a significant impact on N data rates throughput and eco-efficiency administrators will naturally be itching to turn that protection off and operate in what's called Greenfield mode, a mode where we don't have to co-exist with ABG.
Now that's a probably more viable approach in 5 GHz span where fewer exist for legacy coexistence. Although we are talking about a wireless network upgrade there will also be impacts on your upstream distribution network. For most companies that's a wired network where access points are tethered by Ethernets, edge switches and a wire controller. Those elements will be expected by increases in traffic and power consumption. If you were to replace every legacy access point in your network with a brand new N access point application throughput could increase at least in order of magnitude.
However, actually traffic analysis will probably show that doubling the load on your controller and your distribution network is a lot more likely. At least during initial rollout when you are fussing with coexistence and you are not using some of the n options like [frame] [inaudible 19:22] over the next few years that traffic is going to change. As you pile on new users devices and applications you will deploy more access points to add capacity and coverage. Furthermore those legacy clients that operated at 54 mgbs they will get replaced by new N clients that operate at 300 mgbs or higher.
This wireless evolution will clearly increase the demands placed on your wired network. You obviously have to plan for those upgrades as well. What exactly do you plan for? Let's start with one example. Let's examine the load increase on this Ethernet back-haul links between your high throughput access points and edge switches. This slide shows excerpts from the detailed traffic analysis in his paper, The All Wireless Workplace. You might not be able to read the small numbers but Peter starts with a 50% mix of 802.11 arg clients at an average rate of about 36 mgbs. And 802.11n clients using 2 x 2 MIMO at an average rate of 104 mgbs that's g or 216 mgbs, that's 4a.
On a dual radio access point operating in high throughput mix mode in both spans this fields 206 mgbs of traffic. On all of those 802.11 ag clients are retired and replaced by 802.11n clients, still using 2 x 2 MIMO but at a slightly faster data rate. The maximum expected traffic flows through to 400 mgbs. This includes all the protocol overhead of 802.11n and on the Ethernet side there is lower overhead. You are looking at slightly smaller number. At that point you are fast approaching the saturation of the 10100 Ethernet up-link that makes it a traffic bottle neck during peak periods. This is the case for upgrading your 10 100 edge switches to gigabit Ethernet, but that's probably not an immediate need.
the load and [bose] wireless will increase over time in most cases immediate gigabit Ethernet upgrade won't be required. You have to plan for them in the future as your network grows. Further insight into capacity planning, I highly recommend Peter's paper that is referenced here. Of course, there is another wireless component connected to your wireless network and that is your wireless controller. Here product architectures vary widely from fat access points to thin access points to controllers to hybrid architectures that redistribute some functionalities back to the edge again.
This figure illustrates the potential bottleneck that is created by centralized wireless controller that pushes the traffic forwarding deep into the network. The redline shows traffic from wireless clients to somewhere distance perhaps in a data center. In this case there is not a huge difference if you take the wireless controller out of the data path or you don't. However the green line shows the traffic sent by one wireless client to another. The destination could be associated to the same access point or to another access point on the same sight. In this case funneling all wireless traffic through a central controller is inefficient. That controller could easily become a bottleneck for high throughput applications and add unacceptable latency to voice traffic.
To address this problem some end vendors are decentralizing their architectures shifting functions back to the access point splitting controllers into central management plans and data plans uses that can be deployed close or even on access points. Functions like 802.1x authentication and tagging should ideally occur at or near the access point and not somewhere near the core of your network.
Finally, before we finish talking wireless network planning let's think about power. 8023as power overuse Ethernet. In enterprise, wireless LANs this is how most access points are powered today. Unfortunately all those transmit receive change added by the MIMO technology they draw power. An aeronet 1250 making maximum use of end features requires 18 Â½ watts. But a standard POE switchboard maxes out just under 13. That's going to come up short. What can we do? In the long tun we can buy new switches that s for the upcoming 8.023at standard. That's being designed right now to support at least 30 watts per port.
What can you do today with what you have already purchased? The easiest thing to do is to cut back on the end options and [radios] that you plan to use at the start. Power draw actually depends on the access points that you have purchased. There is a minimal configuration that your existing POE supports. Of course if you are investing in you want to get your moneys worth you need to give some serious consideration to alternatives and talk to your vendor. Some have answers but that depends on the brand of the Ethernet switch. If nothing else, you could inject more power or you could even resort to plugging it into ac outlets. The bottom line, understand the power implementations of your high performance network design so you don't go speeding onto the freeway and then run out of gas.
Now let's talk about RS networks. Ap placements and configuration determines coverage and capacity. Some vendors are recommending a 1 for 1 upgrade replacing every ag access point with a brand new n access point. That sounds simple you already have power and capable drops in those spots. But it makes some big assumptions. It assumes that your ag network meets your coverage needs and that those needs won't change very much in the near future. What if you are moving from those isolated public area data wireless networks to a much more ubiquitous wireless voice network. Obviously your needs and your coverage needs change. I highly recommend basing your investment in new equipment insulation costs on an RS network plan that has been designed to meet your current and future requirements. Otherwise the virtual guarantee of the design in the field. If you need to move or add access points after the fact and paying your help desk to trouble shoot quality videos and dropped voice calls.
A well designed network in my view is time well spent. Don't be tempted to skimp on this important step. Designing for n can be tricky. Forget those rules of thumb you learned about abg like roughly circulative cells that cover about 3000 sf with MIMO cells can be jagged effected by obstacles like windows, doors, and people. Reception can be better in spots where signal is lower due to the effects of multi-pass. In fact creating the largest possible cell isn't necessarily the best answer for 802.11 n what you really want are cells that are sized correctly to let access points reach client and clients reach access points at a desired data rate.
Fortunately, there are tools that can help you design a good RS network. You will need to provide more plans and coverage requirements. Remember we started defining our user and application needs. Here is where you will be using those requirements in critical input for a program and predictive planning tools. A planning program models RS coverage based on specified characteristics. The planning program may take a CAD drawing or objects that you enter into the program and let the program predict how RS will be attenuated and reflected in that environment. Planning programs suggest where access points should be placed to satisfy goals like design, user density and minimum data rate. You can usually add later to move access points around after the locations are suggested to see how coverage predictions will change without the expense of doing that in real life.
A site survey program measures RS coverage based on things like perceived signal strength. Signal ration and data rate readings taken on site. During a site survey you stage an access point in a desired location and then you move around the cell with a laptop or PDA taking RS measurements. The site survey programs and compiles input to create a big picture. Comparing that to your specified design mold. I think those two tools are valuable. Predicted planning can help you determine likely access point locations, but planning isn't really a substitute for onsite surveys. Ideally performed both during design and then again for verification. 802.11n is especially important to conduct surveys when objects that effect multi-[path] are present. Don't survey an empty building before the furniture gets delivered or the people sit there.
It is important to associate the access points to measure the impact of multipath. In surveys not just with n clients but with abg end clients so that you can meet design goals for both constituencies. Since 5 GHz cells are generally a bit smaller than 2.4 GHz cells I recommend that you design for 5 GHz even if you don't plan to use it initially.
Finally, during post deployment surveys use representative client devices and applications to measure actual application throughput. That's the real user experience. Keeping in mind that MIMO can result in different performance in each direction.
Although channel planning is part of RS design it requires very careful consideration. Has anyone that deployed a BG network knows, co-channel interference can seriously degrade application performance. During RS design we try to avoid co channel interference by assigning non-overlapping channels to adjacent access points. These figures illustrate there are only three non-overlapping channels here in the U. S. Channel planning has always been easier in a 5 GHz uni-band for 802.11a device as well support either 8 or 12 overlapping channels.
Last year, there were 11 new channels made available in the 5 GHz span, but an access point can only use those channels in conjunction with dynamic frequency selections. That's letting the access point get out of the way with something like a radar starts transmitting on that frequency. The good news is that 802.11n by supporting operation and both bans, that actually gives your access point many more channels to choose from. Furthermore 802.11n access point can support channel bonding. In essence, you take 20 MHz channels and you use them to send twice as much data. This is how end can achieve that top data rate of 600 mgps.
During your RS design, you will need to design which patch of air space to allocate to your new N access data points. That decision will be impacted by channels already used in your locations both by your own access points and your neighbors. It must also take into account the bands and channels actually supported by your Wi-Fi client populations. At the moment for most of us those are largely VG clients operating in the 2.4 GHz band. A problem here is that 2.4 GHz is really a prime real estate. It's in the middle of a very busy neighborhood smothered by Bluetooth, cordless phones, microwave ovens, and external wireless access points. Worse, there are only three non-overlapping channels to start with. It's hard to avoid stepping on an already occupied 20 MHz channel. The odds of finding a clear swatch of air using a 40 MHz are slim.
According to the 802.11 standards is on the 80211 N access point that change channels. The net to text co-channel cause interference from a legacy device. As a result of this, the best way to use N in the ISM band is as a migratory path. A place where your continuous support those legacy clients only until they can be upgraded and moved to the 5 GHz band. On both access points you can accomplish this by turning one radius to 2.4 GHz and another to 5 GHz or channel in the 5 GHz band. When legacy clients go away years from now you can re-tune radius 5 GHz and forget about the mind fields.
5 GHz uniband offers considerably more potential for end. Not only are there more 20 MHz channels to choose from but you are more likely to find more space to use 50 MHz channels for high speed applications. This chart is busy but eyeballing it gives you a feel for channel planning. In the U. S. there are 4 distinct sub bands in the 5 GHz stream used by Wi-Fi, some for indoor and some for outdoor use. Bands in the middle can only be used by access points that support the [BFS] feature I described earlier. Which many legacy access points don't and even some new N access points don't.
In short, don't assume that every access point and every client adapter will support all 5 of these GHz channels. During planning determine which channels are fair game, based on the devices you plan to use and their location. Then let your wireless equipment make the best use of those channels. With N an environmental conditions and interface will cause channels to change. But don't insist on manual channel assignments. Instead turn on features like automatic channel selection and auto power adjustments so that your access points can switch themselves out and overlap a little bit but not too much.
When designing a wireless network many people focus on the physical and link layers. It's important not to be myopic after all the purpose of those layers is to carry the traffic that is being generated by business applications. If you are building a high performance network to support applications you are going to need to prioritize the air time provided by those access points and channels. Most business grade and access points and even some soho access points now support Wi-Fi multi media. WMM is the Wi-Fi alliance certification for products that implement portions of the 802.11e quality service standard. If you plan to run voice or video or same platform with data or you simply plan to use both traditional and voice data applications quality service control is necessary to stop bandwidth from starving other applications. WMM accomplishes this by defining poor access categories. Top priority is given to voice and set application requires both agencies. Right behind that is video since it requires significant but consistent capacity. WMM defines two classes for data, best effort and background. WMM adjusts the way that wireless devices share the channel giving higher priority applications more frequent access to the media.
WMM also uses techniques like priority to give higher priority traffic first shot when inside the access point or client. Besides how you want to apply WMM and how you will map these access classes onto your wired network using VLANs and [deserved] tags. Of course make sure you purchase products that support that kind of tagging preferably as the network edge.
Last, but certainly not least, expect your high performance network to require a certain amount of care and feeding. To be sure wireless management monitoring tools are becoming more sophisticated, more automated, and more tightly integrated with your wireless infrastructure. However, the end will definitely require some upgrades. As companies move from best effort data to mission critical wireless business applications characteristics like manageability, reliability, and problem resolution they are obviously going to become more important. The days of manual firmware upgrades, cold start access point traps and stumbling to find ropes, they are long gone.
Don't invest in high performance hardware without making a commensurate investment in automated management and monitoring tools. In a high performance network things can go wrong quickly in a very big way. You need insight into what is happening as well as what will fix it. Let's start with a wireless controllers and wireless management system. If you are building an enterprise class n network your vendor will be supplying this infrastructure. Don't let that be an after thought. Weak tools can waste your time and impact your bottom line. Evaluate management and monitoring infrastructure when choosing your vendor. Look for features like integration on the front end and with event management and IPS tools on the back end. Make sure your vendor provides you with sufficient visibility.
With N its not good enough to know where your access point is. Now you need to know where the client is as well. Not just now, when you are dealing with the problem but at the time the trouble ticket was generated. We have also already talked about the importance of correctly distributing control functionality to avoid bottlenecks as well as automating operational channel like setting assignments. Other features to look for are things like automatic loop sell and spill over.
For years there has been a debate about integrated versus overlay wireless monitoring systems. Personally element managers and obtrusive manager systems have different objectives and therefore excel at different asks. Whatever your preference, don't chose network gear that you can't support whether it is on the phone or by integrating with a third party wireless IPS. As you upgrade to N you will need to upgrade your wireless IPS. Those abt sensors that you might be using today might have to be replaced with N sensors that can hear and start the new protocol including those N access points that are operating in green field mode. You probably will need more sensors too since you will be putting more traffic in the air and you will probably be expanding your wireless plan coverage area.
Some vendors offer access points that can assign a radius to behave at a dedicated sensor but don't forget to also listen in places where you don't actually want wireless to be used. That probably will be the place where someone is going to put an access point that you won't notice. Today most IPS systems can analyze traffic, not just for security reasons but operationally and performance problems. Look for wireless IPS provides expert analysis into end protocols and performance including location and you will get more for your investment.
Finally upgrade your network administrators toolbox so that your there to support incident investigation and trouble shooting. Analyze new end adapters and decoders because end clients may have different experiences also look for tools that let you mimic like voice over and expert that you are going to actually have to trouble shoot it on location. Clearly the more tools that you have that can tell you about what they are actually hearing the faster you will be able to resolve interoperability and code system problems. You don't want the weed through 77 ncs combinations and dozens of channels manually. Find tools that can do that for you.
I'll wrap up here by saying that I am excited about the potential of 802.11n and what I am seeing in terms of maturity in next generation wireless LAN solutions. I think the systems now have the opportunity to make better use of wireless connectivity, broader range of activities and a larger number of users. 802.11n isn't completely responsible for that but it certainly can make a big contribution by extending coverage and capacity and letting clients overcome RS problems that made some older networks unreliable. Simply replacing your legacy access points with N access points isn't likely to create a reliable, fast network.
High performance networks require planning. Traffic engineering will be needed to achieve optimal results without overbuilding or over spending. Indeed when it becomes to performance installing too many access points can be as bad as installing too few. Don't skip over that needs assessment. Establish your expectations and take the time to learn what the quality of service can and cannot do. Then build a network that is carefully designed to meet your current and future needs.
If you lack the tools and skills to accomplish this get a third party to help you. Realize that isn't about access points and controllers. There are two sides to every dialogue. Start upgrading those client devices as well as the wireless tools and features you are going to use to support them. Back to you Amy.
Amy Kucharik: hank you Lisa. That was a great presentation. Before we finish up the webcast I would like to ask you a few questions. Our first question is how do 802.11n high throughput upgrades impact wireless band security?
Lisa Phifer: Like ABG 802.11n can use security measures that are specified by another standard 802.1I because the Wi-Fi allows that requires all Wi-Fi certified products or WPA version 2 that is the certification program for 802.11 I this robust level of security is the minimum you should expect from any 8021 n product. However, 802.11n does introduce new security threats. Specifically because 802.11n clients are more sensitive than their ABT counterparts there are more likely to with external access points. Public spaces will also pose greater risk. A hacker in your parking lot could hear your ABT access points but not form associations. Someone using n might now be able to do some damage. 802.11n products will probably also carry vulnerability. I know vendors are getting better at this but any new product has a chance of harboring a few undiscovered bugs that will need to be fixed with the shear complexity of 802.11n and option make it likely that previously untested scenarios will be found through closing a task.
Finally wireless systems already talked about them are going to have a harder time keeping their eye on. Not only do they have to cover more challenges and greater distances and transit theme forming techniques. In the direction that is helpful to a given client but possible for a given wireless ips sensor.
Amy Kucharik: What about the so called 4th generation wireless LANs. How do they differ from 802.11 solutions.
Lisa Phifer: Some systems refer to products that use channel architectures as 4th generation wireless LANs. I think that is a bit of a misnomer since single channel span isn't brand new and it's not likely to replace more common micro approach. Now what I mean by that adjacent to 802.11 access points usually are set to different frequency to avoid channel interface. Access points can also use things like automatic channel selection to monitor and choose the channel that appears to be unoccupied. But at the end of the day access channels that use 40 MHz channels make that worse. Now single channel architectures, what they do is they tune all access points to the same frequency. Instead of creating non-overlapping cells those products try to create one very large virtual cell by using the wireless to coordinate air time. It's easier because you don't have to find a non-overlapping channel and roam times are reduced because clients don't have to scan around the ban. While those arguments are compelling those architectures are proprietary. That means they only work in homogenous wireless LANs.
First one I think fourth generation of wireless LANs will be the those that incorporate other advances. Advances like transmit beam forming. Like hiding set drive antennas and wireless mesh distribution systems. Those standard base solutions can sit more easily into an existing or multi vendor network but today make better use of data to 8011 n than your average dual radio access point does.
Amy Kucharik: In your presentation you recommended verifying performance by measuring applications throughput are there any open source tools that you recommend using for that?
Lisa Phifer: I get that asked a lot. During site surveys, many people use familiar programs to spot check performance. For example, finger trace routes can be used to get a quick feel for latency. You can stt a large file several times to get a sense of both data throughput. Alternatively some wireless clients have built in performance test panels that can be used to exchange a specified number of data frames with access points that come from the same vendor. If you are looking for a test program that is automated but can measure end to end application throughput, then you probably want a tool like [Iper]. Iper is a popular TCP and EDP benchmarking program that reports on bandwidth, delaying, jitter and data gram lost between a pair of hosts. You can download Iper from the open source clip like [floor storage]. You launch one copy in a server mode on a laptop connected to your access point preferably by Ethernet. Then you launch another copy in client mode on the laptop or PDA being used to conduct your site survey. Command line parameters can help you simulate different application payloads.
Amy Kucharik: One more question. As companies deploy 802.11n what steps should they take ensure network scalability?
Lisa Phifer: Probably the simple most important step you can take is to maximize your use of the 5 GHz stand. Even if you need to support 2.4 GHz clients. Today you will want to migrate the 5ghz as you lay the path for the new users and new applications. If you place your access points where coverage is good for 5 GHz and you make sure you buy access points that support 5 GHz variation, then you will be laying the ground work for that future migration.
Some people recommend overbuilding your wireless LAN so you have the capacity for your growth. I think you can accomplish this benefit by planning you wireless to provide more capacity by deploying that plan in phases. For example, if you are creating a wireless today consider planning access point placement in a way that it would provide seamless coverage you would need if you moved to voice. Voice networks demands more than redundancy. That kind of stuff can be very expensive to back fill. If you know where your access points will be needed for voice you can then deploy some set of those access points today where you need them for data. A planning tool can help you thin out your access point count in that fashion.
Amy Kucharik: Thank you Lisa. I appreciate your responses. It was a pleasure talking to you today. Before we end did you have any final thoughts or comments that you would like to share with out audience.
Lisa Phifer: With ratification of 802.11n later this year, our industry will have reached an important turning point moving from casual and largely unplanned wireless active access to a secure wireless networks. Making that transition is going to take some careful planning in network engineering to make sure our business are understood and met with a high degree of reliability. My hope is today's presentation will help our audience put together a solid game plan for their own network and their own next generation high performance wireless plan.
Amy Kucharik: Thank you. That concludes today's online event.