Hacker News Reader

Comment by jread | Hacker News Reader

If you are interested in DNS performance comparison, I created a tool for comparing DNS query times of 16 different hosted DNS providers including Route 53, Dyn, DNS Made Easy, UltraDNS, Rackspace, SoftLayer, Easy DNS, Zerigo and a few others here: http://cloudharmony.com/dnstest

It is a browser/javascript based test that uses a wildcard DNS record and alternating between cached and non cached lookups to determine an approximate query time for each provider.

View on HN

I ultimately went with Route 53 because of the performance. Glad to see your tests confirms my choice.

Switching to Route 53 from MediaTemple as my DNS was like a night and day difference. The site felt much faster.

I'm a little confused here. Wouldn't client DNS resolution performance really depend on your local DNS cache (e.g. the DNS server at your ISP)? Nobody going to your site will hit Route 53 or MediaTemple's DNS servers directly. These only serve as the source of truth for the thousands of DNS caches out there that clients actually hit.

Typically this is true as the only performance hit will be due to the initial lookup, and then your resolver will cache the record. However, if your DNS record uses a low TTL (required for things like DNS-based monitoring/failover) like 30 secs or 1 minute, DNS lookups will have a larger impact on site performance because user resolvers will have to frequently re-query the authoritative DNS.

Providers that utilize an Anycast DNS infrastructure (Route 53, Dyn, DNS Made Easy, Cotendo, Easy DNS) will generally be faster for lookups than those that do not due to lower network latency. This is especially true for international users where latency might be 500ms higher than in the US.

Reply on news.ycombinator.com