[178162] in North American Network Operators' Group
Re: OT - Small DNS "appliances" for remote offices.
daemon@ATHENA.MIT.EDU (Maxwell Cole)
Wed Feb 18 16:07:18 2015
X-Original-To: nanog@nanog.org
Date: Wed, 18 Feb 2015 10:21:42 -0500
From: Maxwell Cole <mcole.mailinglists@gmail.com>
To: "nanog@nanog.org >> 'NANOG list'" <nanog@nanog.org>
In-Reply-To: <54E4A9BF.5000006@alter3d.ca>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
+1 for the pi,
The new model has a quad core and 1GB of ram which should be more than
enough for a DNS.
On 2/18/15 10:03 AM, Peter Kristolaitis wrote:
> Not "industrial grade", but Raspberry Pis are pretty great for this
> kind of low-horsepower application. Throw 2 at each site for
> redundancy and you have a low-powered, physically small, cheap, dead
> silent, easily replaceable system for ~$150 per site. Same idea as
> the Soekris -- just ship out replacements instead of trying to repair
> -- but even cheaper.
>
> Between having 2 (or more) at each site, plus cross-site redundancy
> via anycast, it would be pretty robust (and cheap enough that you
> could have cold-spares at each site).
>
>
>
> On 02/18/2015 09:28 AM, Ray Van Dolson wrote:
>> Hopefully not too far off topic for this list.
>>
>> Am looking for options to deploy DNS caching resolvers at remote
>> locations where there may only be minimal infrastructure (FW and Cisco
>> equipment) and limited options for installing a noisier, more power
>> hugnry servers or appliances from a vendor. Stuff like Infoblox is
>> too expensive.
>>
>> We're BIND-based and leaning to stick that way, but open to other
>> options if they present themselves.
>>
>> Am considering the Soekris net6501-50. I can dump a Linux image on
>> there with our DNS config, indudstrial grade design, and OK
>> performance. If the thing fails, clients will hopefully not notice due
>> to anycast which will just hit another DNS server somewhere else on the
>> network albeit with additional latency. We ship out a replacement
>> device rather than mucking with trying to repair.
>>
>> There's also stuff like this[1] which probably gives me more horsepower
>> on my CPU, but maybe not as reliable.
>>
>> Maybe I'm overengineering this. What do others do at smaller remote
>> sites? Also considering putting resolvers only at "hub" locations in
>> our MPLS network based on some latency-based radius.
>>
>> Ray
>>
>> [1] http://www.newegg.com/Mini-Booksize-Barebone-PCs/SubCategory/ID-309
>
>
