Boarding House Wireless - Meru vs. Xirrus

[Jona]

Hi Edugeek,

We're looking a number of wifi solution to provide wifi coverage to the approximatly 600 students in our boarding houses.

 

Obviously coverage is a key issue for us as the boarding houses are much more like hotels in construction than classrooms e.g. lots of little rooms spread over several floors. At least one boarding house is a classic old building with big thick walls.

 

We have a network access controller (Bradford campus manger) which we need to work with the wireless.

 

The front runners currently seem to be Meru and Xirrus, would be especially interested to hear any feedback good or bad about these solutions, especially if your using a Campus Manager.

 

Thanks

Jona

[Dragon]

Don't forget Ruckus!

[m25man]

How much bandwidth do you have to share with your 600 boarders?

Throwing a £20k system at a bunch of boarders that only have limited Internet capability seems a bit of an overkill to me!

 

If you need to service all 600 users from the same few AP's your looking in he right place however if your only going to have a dozen or so clients attached to each AP sharing a 100 Mbps leased line the likes of a 16 radio Xirrus unit is possibly a bit over specified?

 

Think very carefully about how you are actually going to use it, trial it first before committing to it.

[glennda]

Been playing with Meraki over the last few days seems really good. The controller is cloud based and therefore pretty much plug and play (getting config from the internet).

 

Access points are also fairly robust with decent signals.

[glennda]

Have you considered Aerohive? Or Meraki like Glennda said if you're hankering after a controller-based solution (though having a controller in the cloud does seem risky- maybe a more traditional vendor like Ruckus????)

Meru's share prices are a bit shakey at the moment, and the way Xirrus ties all their radios together usually works better in big open areas?

 

Theres nothing wrong with having the controller in the cloud. I've got higher end NFR units arriving tomorrow and from my conversations with the tech guys at meraki nothing leaves your network and the controller is purely there for managing the AP's.

Edited May 28, 2012 by glennda

[joshbrowne]

Theres nothing wrong with having the controller in the cloud. I've got higher end NFR units arriving tomorrow and from my conversations with the tech guys at meraki nothing leaves your network and the controller is purely there for managing the AP's.

 

But then why do they have a controller at all? From my research you lose at least some of the functionality of the network- for example, if CWP is stored on the Meraki server and you lose connection, obviously you cannot authenticate users that way. I would argue that the features that aren't lost probably aren't there anyway, i.e. , L3 roaming, load balancing etc. With regards to local data traffic, if access to the controller is lost, local forwarding only works for those already connected. If they then try to roam, they'll fail, the same goes for if they try to reconnect.

 

If you're going for a controller based solution, it's better to go for a physical device on site IMO, it's less likely to go down that your net pipe.

[Haptic]

I have PM'd.

 

Chris

[rbance]

Get a survey done....I suspect Xirrus will do this very well....

Depending on what you have, one Xirrus AP could do 600 devices easy.....but then things like geography (coverage) play a big part which may cause an issue for you....and you may find a solution like Meru or Ruckus may work better....

 

What is your backend like? 1 GB ? 10GB? you may need to take into account the load on the network and the types of traffic you are expecting as well...if 100MB then web traffic would be viable but not much else...

 

 

 

For my situation, Meru and Ruckus were not close....

[Haptic]

it's worth mentioning that the new Ruckus 7982 AP's can handle extremely high density deployments i.e. 500 concurrent connections, plus when coupled with FlexMaster is ideal for multisite deployments for the full details have a look at Ruckus brings out beast

[SuperfluousAdjective]

But then why do they have a controller at all? From my research you lose at least some of the functionality of the network- for example, if CWP is stored on the Meraki server and you lose connection, obviously you cannot authenticate users that way. I would argue that the features that aren't lost probably aren't there anyway, i.e. , L3 roaming, load balancing etc. With regards to local data traffic, if access to the controller is lost, local forwarding only works for those already connected. If they then try to roam, they'll fail, the same goes for if they try to reconnect.

 

If you're going for a controller based solution, it's better to go for a physical device on site IMO, it's less likely to go down that your net pipe.

 

The controller adds functionality and an interface to manage the WAPs. The APs themselves have enough RAM to store their configuration so if the link to the controller goes down very little functionality is lost until it comes back up. You just can't make changes to the configuration of the network or use their online services. But they have redundant data centers across the globe.

 

Keeping controllers in house or in the cloud will be a long debate for any facet of networking. But more and more applications have been moving to the cloud such as Google Apps for Education, Live @ Edu. If you had your own internal mail server you could keep mail services up for your organization, but the idea of managing an extra server, backing it up locally, managing that, power, etc.. it is often easier and cheaper to just do it in the cloud. In many cases it also adds functionality to what you could have internally because it is a shared resource and shared funding. Plus it saves you a lot of headache. Who wants to manage and maintain an Exchange server? That said, a wireless controller is something I would happily put in the cloud.

[SYNACK]

The controller adds functionality and an interface to manage the WAPs. The APs themselves have enough RAM to store their configuration so if the link to the controller goes down very little functionality is lost until it comes back up. You just can't make changes to the configuration of the network or use their online services. But they have redundant data centers across the globe.

 

Keeping controllers in house or in the cloud will be a long debate for any facet of networking. But more and more applications have been moving to the cloud such as Google Apps for Education, Live @ Edu. If you had your own internal mail server you could keep mail services up for your organization, but the idea of managing an extra server, backing it up locally, managing that, power, etc.. it is often easier and cheaper to just do it in the cloud. In many cases it also adds functionality to what you could have internally because it is a shared resource and shared funding. Plus it saves you a lot of headache. Who wants to manage and maintain an Exchange server? That said, a wireless controller is something I would happily put in the cloud.

 

All the intelligence is stored offsite, all the redundant datacenters in the world mean nothing if your internet link goes down. Sure people already on can stay on but anyone turning their laptop on after is stuffed. Its mad putting basic control stuff outside the environment it services. Its like putting all your power switches for appliances in a room on the other side of the country.

[SuperfluousAdjective]

All the intelligence is stored offsite, all the redundant datacenters in the world mean nothing if your internet link goes down. Sure people already on can stay on but anyone turning their laptop on after is stuffed. Its mad putting basic control stuff outside the environment it services. Its like putting all your power switches for appliances in a room on the other side of the country.

 

This is false. The cloud controller just sends updates of network policy changes to the WAPs. Users continue to stay connected, new users can authenticate, users can roam between WAPs, layer 7 firewall policies/traffic shaping on the WAPs continues to run, and VPNs continuing running.

 

The data traffic is processed internally in the WAP. The cloud provides a management interface and a means for the APs to get updates. If you have an outage at a data center some services in the cloud controller may be done temporarily, but everything critical continues to run on site as it was before the controller went down. Your local controller (if no HA unit) going down would cause far greater issues and far more downtime. Whereas with Meraki your data is replicated within the data center itself and then to another data center as well. Less opportunity for downtime. If your internet connection goes down you're out of luck anyway. Local traffic and authentication will continue to operate. We haven't had an outage with Meraki, but I have done testing while setting up ACLs on the VLANs and it works like a champ without access to the cloud controller. You can't tell the difference except for the lack of internet, which you'd have anyway.

 

Most people don't have redundant wireless controllers anyway so the point of failure is greater. I build my network to be as redundant as possible, but if there is an ISP outage the last thing I'm worried about is creating new wireless policies.

Edited September 11, 2012 by SuperfluousAdjective

[SYNACK]

This is false. The cloud controller just sends updates of network policy changes to the WAPs. Users continue to stay connected, new users can authenticate, users can roam between WAPs, layer 7 firewall policies/traffic shaping on the WAPs continues to run, and VPNs continuing running.

 

The data traffic is processed internally in the WAP. The cloud provides a management interface and a means for the APs to get updates. If you have an outage at a data center some services in the cloud controller may be done temporarily, but everything critical continues to run on site as it was before the controller went down. Your local controller (if no HA unit) going down would cause far greater issues and far more downtime. Whereas with Meraki your data is replicated within the data center itself and then to another data center as well. Less opportunity for downtime. If your internet connection goes down you're out of luck anyway. Local traffic and authentication will continue to operate. We haven't had an outage with Meraki, but I have done testing while setting up ACLs on the VLANs and it works like a champ without access to the cloud controller. You can't tell the difference except for the lack of internet, which you'd have anyway.

 

In which case the tag line is a tad misleading, if it runs things locally then it is actually controlled locally, just configured through a remote interface. At that point what is the point apart from bleeding you for a subscription to a website.

 

Why not just go with something like the Aruba IAP series which virtualizes the controller inside the APs themselves. It just elects one as the master and the rest follow, if it goes down another one takes over being the master. As long as you have one AP up it is managed and you can configure anything with full management. Way more resilient as the only way you have no management is if you have no APs, at that point what does it matter.

 

Even the Ruckus APs will still keep running without the controller and authenticate via static means, its not a massively special feature.

[SuperfluousAdjective]

In which case the tag line is a tad misleading, if it runs things locally then it is actually controlled locally, just configured through a remote interface. At that point what is the point apart from bleeding you for a subscription to a website.

 

Why not just go with something like the Aruba IAP series which virtualizes the controller inside the APs themselves. It just elects one as the master and the rest follow, if it goes down another one takes over being the master. As long as you have one AP up it is managed and you can configure anything with full management. Way more resilient as the only way you have no management is if you have no APs, at that point what does it matter.

 

Even the Ruckus APs will still keep running without the controller and authenticate via static means, its not a massively special feature.

 

The Aruba IAP series is fine if that is what you are looking for. The lack of advanced features and 16 AP limit is a huge hindrance for us and takes the product line out of the enterprise market and into the small/medium business market as far as I'm concerned. Your entire network is also limited to the CPU of a single access point (with a 16 AP limit which tells you there are problems there). Ultimately if you want a true enterprise wireless network you have to buy a controller from them. We actually received a bid from Aruba and they came out considerably higher for their mid-level access points compared to the top level Merakis we put in.

 

Meraki offers pretty steep educational discounts (think 50% or more). The controller/license fee includes warranty and support.. 24/7/365 next day shipment of any failed product. Their controller/license/warranty fee was actually cheaper than a lot other bids we received that had the added cost of a controller (only one, no redundancy). The cloud controller also includes a lot of great reporting/monitoring features common in other controllers as well as Apple mobile device management and some other really useful features we make use of in our network. We didn't choose Meraki because they had a cloud based controller, although it was a nice to have. They have some interesting features in their cloud controller that we make use of. They performed well in our tests and their price was extremely competitive. We don't expect to have any wireless system for more than 5 years so investing in a controller doesn't seem to make a lot of sense. With these it is just a 1-1 swap. The Cat6 lines are run, the mounts are installed. APs hung. When we want to do a refresh we just pull the old down and upgrade to the new... preferably again without a controller. With the way networking is progressing I can't imagine wireless controllers will be around much longer. There isn't much point to it. Wireless management will move to the cloud, virtual appliances or they'll become internal. But the days of buying a hunkin box to manage access points are numbered.

 

I never said running without connection to a controller was a massively special feature. I was just saying that you were wrong when you said you can't authenticate to the network if a Meraki AP loses its connection to the controller. That's false. Nowhere did I state that Aruba or Ruckus did not provide useful features or a fine product. Although, after you brought up the Aruba Instant AP line I do stand by my statement that it is meant more for a small branch office and not an enterprise environment.

Edited September 11, 2012 by SuperfluousAdjective

[SYNACK]

You are right the IAP series does have its limits, it does sound like the cloud ones may have some of the same issues you were talking about though, if they are disconnected from the controller but still run the featureset surely they are doing something very similar to the IAP series so may be subject to some of the same overheads with CPU and resources.

[SuperfluousAdjective]

You are right the IAP series does have its limits, it does sound like the cloud ones may have some of the same issues you were talking about though, if they are disconnected from the controller but still run the featureset surely they are doing something very similar to the IAP series so may be subject to some of the same overheads with CPU and resources.

Not exactly because each AP is only responsible for itself and some functions are managed by the controller, they will just continue to operate as is until new instructions are given. Things like channel selection and power changes for interferance management is down while no access to the controller. And since each AP is responsible only for itself it becomes a distributed overhead. Whereas with Aruba there is one AP assigned to both control and manage all other APs in the network, albeit with a silly limit of 16. There is no distribution of resources. This is similar to what Aerohive tried in the past. They killed the program because it sucked. The feature set you can provide is lacking and the limit of the number of APs is annoying and expensive when you hit the wall. Turns out it is just more beneficial all around to put a controller in the cloud or distribute the management among the APs.

[dave.81]

We are a boarding school, although much smaller, using Ruckus. We have some very old buildings and have had to use more AP's in some areas than we had hoped due to the building style. We are just looking at buying more AP's to cover a few dead spots.

 

As a whole the system has been great although the lack of support for our current controller has been a sore subject seen it's only 3 years old and no longer relieves updates.

[Haptic]

We are a boarding school, although much smaller, using Ruckus...As a whole the system has been great although the lack of support for our current controller has been a sore subject seen it's only 3 years old and no longer relieves updates.

 

Hi Dave, I'll drop you a PM as we might be able to do something on the 1000 to 1100 upgrade.

 

Thanks,

 

Mark

[jc666777]

Not exactly because each AP is only responsible for itself and some functions are managed by the controller, they will just continue to operate as is until new instructions are given. Things like channel selection and power changes for interferance management is down while no access to the controller. And since each AP is responsible only for itself it becomes a distributed overhead. Whereas with Aruba there is one AP assigned to both control and manage all other APs in the network, albeit with a silly limit of 16. There is no distribution of resources. This is similar to what Aerohive tried in the past. They killed the program because it sucked. The feature set you can provide is lacking and the limit of the number of APs is annoying and expensive when you hit the wall. Turns out it is just more beneficial all around to put a controller in the cloud or distribute the management among the APs.

 

The Aruba IAP system is actually unlimited as of early this year. The controller AP is just there if you want to change something it doesn't do any co-ordination of any other APs. If it dies another AP assumes the 'management' IP address.

 

So there is distribution of function with aruba

 

Other good aruba feature all the APs are lifetime warranty and you don't pay any support feed but with meraki if you stop paying your yearly fee you wont be able to manage your system any more.

 

So I would argue look at the 5 year costs.

 

DISCLOSURE I work for Aruba.

[SuperfluousAdjective]

The Aruba IAP system is actually unlimited as of early this year. The controller AP is just there if you want to change something it doesn't do any co-ordination of any other APs. If it dies another AP assumes the 'management' IP address.

 

So there is distribution of function with aruba

 

Other good aruba feature all the APs are lifetime warranty and you don't pay any support feed but with meraki if you stop paying your yearly fee you wont be able to manage your system any more.

 

So I would argue look at the 5 year costs.

 

DISCLOSURE I work for Aruba.

 

Thank you for the update. I wasn't aware of the change. I'll have to look back at your product offering for the next wireless build-out we do and see how it compares to the Aerohive solution. I personally prefer a distributed or cloud based controller solution. As far as the five year costs we recently received many quotes on a wireless deployment of about 200 APs from many vendors. An Aruba bid was made using AP105 access points and a controller. Five year cost was about 35% more expensive than the comparable Meraki MR16. This obviously is dependent upon region, hardware vendor, level of educational discount and many other factors so I assume the price differential we received is not the same across the board. But for us the difference we saw was very great. Even if you fully removed the Aruba controller from the quote the 10 year cost (assuming pricing doesn't change) on Meraki would be cheaper than the 5 year cost of Aruba (with no controller costs). They are pushing some pretty heavy educational discounts around here. A lot of districts in this area have been going with them lately. We called about 6 of them before deciding on Meraki and all of them were very pleased with the product. Not a single negative statement other than one client who wasn't using key caching would have liked to see better roaming performance.

Edited September 15, 2012 by SuperfluousAdjective

[SYNACK]

Thank you for the update. I wasn't aware of the change. I'll have to look back at your product offering for the next wireless build-out we do and see how it compares to the Aerohive solution. I personally prefer a distributed or cloud based controller solution. As far as the five year costs we recently received many quotes on a wireless deployment of about 200 APs from many vendors. An Aruba bid was made using AP105 access points and a controller. Five year cost was about 35% more expensive than the comparable Meraki MR16. This obviously is dependent upon region, hardware vendor, level of educational discount and many other factors so I assume the price differential we received is not the same across the board. But for us the difference we saw was very great. Even if you fully removed the Aruba controller from the quote the 10 year cost (assuming pricing doesn't change) on Meraki would be cheaper than the 5 year cost of Aruba (with no controller costs). They are pushing some pretty heavy educational discounts around here. A lot of districts in this area have been going with them lately. We called about 6 of them before deciding on Meraki and all of them were very pleased with the product. Not a single negative statement other than one client who wasn't using key caching would have liked to see better roaming performance.

 

That does make sense though, a lot of the products are priced for the US market and are rubbish value in any other market. That is how we end up with stuff that would be the more expensive stuff instead as when you apply the you're not American, up your rear tax to 'cheaper' stuff it are horrifically overpriced. It is actually cheaper to buy stuff in the US and ship it here then retrofit it with local PSUs than it is to buy it locally thanks to the insane premium that is applied to anything outside the states. Lots of stuff that makes perfect sense in the states makes no sense anywhere else in the world, like Apple products.

Edited September 17, 2012 by SYNACK