Recommended wireless AP for 400 users office

Sean Hunter jamesb2147 at gmail.com
Wed Feb 4 02:30:39 UTC 2015


I happen to administer a deployment of almost exclusively Meraki gear; ~140
switches (mix of MS42 and MS22) and ~400 AP's (almost all MR16's).

I would *not* recommend them for this situation. If you've got a low-usage
scenario, they might be fine. The tech support quality has noticeably
declined over the last 2 years we've been running their gear, and the
really amazing fact about that is that I'm working with the same people
(read: Cisco is making them script-following support monkeys rather than
techs) who generally know me by name. That is another "interesting" point
with Meraki. I've helped them identify several bugs, some of which were
very serious. We regularly have to ship back switches after an update.
We've encountered a RADIUS auth issue where users were being randomly
diverted into the wrong VLAN in the middle of a wireless session (they
weren't even roaming or anything). The RADIUS issue was actually really
interesting; it dumped users into our management VLAN which very quickly
depleted the DHCP pool. About 20% of our 4000 wireless devices were in the
wrong VLAN and unable to get on the internet (yikes!) and suddenly our AP's
started bouncing because they lost their DHCP leases, couldn't get new
ones, lost contact with the Meraki cloud controller, and started rebooting
every few minutes (the MR16's don't boot quickly, either). It was
terrifying and horrible, especially because that was the 2nd time it
occurred for us. We're *still* running a custom Meraki firmware that's a
year old because they have, twice now, reported that the fixed the RADIUS
issue, only to have us experience this when we updated them all at once.
We've had similarly critical firmware regressions on the wired side of
things, aside from the normal slew of issues ("What do you mean your
firmware upgrade disabled the uplink port?").

If provided a do-over, I'd select Ubiquiti today, or another of the more
professional vendors. Meraki's gear is cool, the Dashboard is a *dream* to
work with, I love the built-in remote packet captures, and they're probably
fine for most small deployments, but Meraki isn't ready for prime time yet.
I feel like a beta-tester rather than a customer, and the support is
getting worse when, if they're going to act like a startup (read: move fast
and break things), they really need for it to get better.

RE: Aforementioned criticisms from this thread:

1) Meraki makes you buy hardware, licenses, and more hardware when the
first dies.

Response: Almost 100% wrong. I read each warranty and suggest you do the
same for any gear you buy. The stuff we use (MR16's, MR22's, and MR42's)
has cost-free replacement warranty coverage as long as you hold a valid
license. The one exception are the outdoor AP's, which only have a 1 year
warranty, which is rather crappy on Meraki's part, because the license fees
are the same no matter your model of AP (indoor, outdoor, big and
expensive, or small and cheap).

2) Meraki switching/AP functionality is/is not tied to cloud controller
functionality.

Response: It is and it isn't. First, you must have a valid license or 30
days later your network ceases to function. All of it. Completely ceases.
They haven't been flexible on this and we even got within 2 days of it
expiring when we first installed ours. Our sales rep was sympathetic but
unhelpful, even after taking our money for the license. :/ Second, we've
had our cloud controller go down and life went on. However, we've also had
our AP's be unable to get a DHCP lease and start rebooting every few
minutes. You tell me what that's worth.

I think that might be $0.05 worth. ;)

On Mon, Feb 2, 2015 at 4:24 AM, Tim Franklin <tim at pelican.org> wrote:

> > That's it. Step 1, buy the equipment at full price. Step 2, pay for the
> cloud
> > management license, yearly. Step 3, no extended warranty option, so pay
> full
> > price if equipment from step one fails.
>
> As long as you're doing step 2 (which you *have* to, otherwise it's a
> brick), isn't step 3 "report device as failed, new device shipped to site,
> plug in cable, sucks down config of old device from the cloud, up and
> running again"?
>
> I only so far have the demo gear from one of their (rather good) training
> courses, which has a couple of years left to run, rather than any live
> deployments, but that's my understanding of the support model from the
> meetings I've had with them to date.
>
> Regards,
> Tim.
>



More information about the NANOG mailing list