Using dynamic DNS and TTL as a poor man's redundancy

[Kevin Darcy]

Charles Cala wrote:

>--- Rick Kunkel <NOSPAM-kunkel at w-link.net> wrote:
>
>  
>
>>He doesn't want these servers hit during under normal
>>operating
>>circumstances.  They're for failover only.
>>    
>>
>
>  
>
>>In essence, we'd be
>>running a dynamic DNS server with a very low TTL 
>>    
>>
>
>  reasons why dns gslb has issues
>http://www.simplefailover.com/outbox/dns-caching.pdf
>http://www.tenereillo.com/GSLBPageOfShameII.htm
>http://tenereillo.com/BrowserDNSCache.htm
>
>basic overview of a "routed" fix.
>http://www.quickeagle.com/pdf/infocenter/Safety_in_Numbers_WP_Final.pdf
>each has its own issues, problems, and solutions.
>
>
>ya may want to look at http://www.akamai.com/index_noflash.html
>but they have "burps" as well.
>http://www.theregister.co.uk/2004/06/15/akamai_goes_postal/
>  
>
Having said all of that, though, many folks use DNS-based GSLB and it 
works fairly well for them/us. The thing that gets me, though, is the 
moral aspect of DNS-based GSLB, i.e. low TTLs don't just waste the 
resources of the authoritative servers for the relevant zone (and 
associated network bandwidth, etc.), they also waste the resources of 
every resolver trying to resolve the name(s) in question. It's like a 
drunk driver on the road, not just endangering himself/herself, but also 
everyone in the immediate vicinity. It's like a moocher of other 
people's resolver capacity; what I think economists call "externalizing 
costs".

That's why DNS-based GSLB leaves a bad taste in my mouth; unfortunately, 
I didn't have any say in whether we should use it here or not...

                                                                         
                                                   - Kevin