New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Yet*
An Asia POP is 2nd on the "most requested feature" next to better documentation/examples.
We'll most likely open an Asia location later this year, even if it's just for anycast customers.
Thanks for the interest,
Francisco
So, how critical is DNSSEC to have really?
Interesting. However, in my case, one or all approach is not enough. I'm working on a custom CP making it possible to create or modify zones based on: (1) domain category or (2) domain names, using wildcards. Read: script will create/modify zones selectively based on above (1) or (2). Any other ideas on mass zone management always welcome
I am not qualified to answer that question.
With gDNSd the zone record can have an IP number for a host name that is not geocast, and/or a statement like "geoip!mysvrpool" in place of the IP for geocast host names. The "geoip!mysvrpool" is defined in the config file with IPs for the geocast server pool (datacenters). If you were going to change one of the servers in the pool and were using BIND you would have to change the IP in each zone/view for all the domains affected, with gdnsd you would only have to change it once in the main config file.
With BIND you would need to have and change a zone file in each view, more views more files to change, even if that domain did not have any geocast hosts. With gdnsd there is only one zone file. I would think that this would make automating maintenance a lot easier.
I am not sure exactly what you are looking for in a dns panel, but after you get set up and know what you want based on gdnsd, PM me if you don't mind a responsive panel written in perl, since I will be making one for myself anyway.
Lowend anycast/geocast DNS services. Thank you @Francisco @vld and @adamBB for your assistance in making this happen
Why don't you use Rage4?
Why the fuck am I getting notifications for someone commenting on this thread I HAVE HAD NO INVOLVEMENT UNTIL NOW ARRRRRGH
Favourited it (Those stars on the homepage)?
I'm wondering what the point of Geo DNS is when the servers they point to aren't geo distributed, too - which quite probably most aren't.
Once an operation is big enough it will either buy that as service anyway - the whole packet.
That said I find it a shame that most DNS servers still do not offer solutions for that task (I wouldn't use bind even if I were given super-beautiful nymphomaniac women for it and PowerDNS, oh well ...)
Wat?
@aglodek
My point was: a) there is no official mechanism for providing geo distributed authoritative NS servers. And even if there were, say, 4 NS servers for a domain and they were smartly spreaded over the planet it would still be lottery which one is contacted. Moreover it'd be lottery again which one the recursors would use and which target related response they would provide.
b) What we really want is the "payload protocol/service" to be geodistributed, e.g. our web-shop.
c) For all of that to work pretty every site (assuming not purely static content) would run into other sets of problems, e.g. DB related.
Well noted, I consider geo distribution a good thing (which is why I complain about just few DNS servers supporting it) but usually it's reasonably used for major projects and the complexity is far higher than "use Geo DNS and be done".
Sending the client to a smartly selected host is just the first step (and usually the easiest to solve problem).
Furthermore, we have more urgent problems with DNS. For instance the security question. And DNS still being an excellent attack amplifier hints a major practical issues (like not at all being idiot proof while shockingly often being admin'd by idiots).
@bsdguy - You are completely right, if I have one web server with some websites on it.
But lets say that I have customers all over the world or a large part of it and I need to offer services in four or five or more languages and I want them to see the page in their own language when they come to the site. Geo locating the client to the US or Chile will get me a default language that is closer to correct then just sending everyone to an English site. Think google news.
Now if I can anycast dns and get your dns request to the closest dns server, and geo locate you to the closest web server, say with 100ms spacing. You will most likely be happier with the services if they are fast and in you own language. Would it be better if I just offered a slow English website in Amsterdam and you are from Chile.
I did consider just having different domains on different web servers in different locals, in different languages, but wheres the fun in that.
By the way I have used BIND for years and they never gave me super-beautiful nymphomaniac women, but it would have been a big plus if they did.
Good to know.
Can't afford it long term,
EDIT: For me it would be about $40 a month for RAGE4 DNS
Uhm, sorry no. You can simply make use of http/Browsers to get that done. In fact Browsers even tell your server the users preferred language.
I don't mention that because I'm picky or evil-minded but because it demonstrates part of the problem. There are many factors to what may seem to be 1 problem but actually is a group of (often related) issues.
I get what you mean. But there's still some cans of worms waiting for you. A classical one being DBs.
Again, I'm certainly not against geo distribution. I'm just suggesting that one analyses properly and designs/buys a comprehensive and well thought out and balanced solution. Sometimes the result of such analysis should reasonably be: "Not worth it" or "Too complicated for our needs".
See! That's what I say. How much more proof can one ask to see that bind is EVIL! **g
@bsdguy - What's a DBs?
DBs ~ Databases.
I don't like them and I don't use them. They are Evil.