We've been running SafariNow with a subdomain wild card DNS record for some time. We supported some integrations that allowed partners to link using subdomains for e.g. getaway.safarinow.com.
So we implemented the wildcard DNS record.
Sorted. Except we recently discontinued this integration, but we continued seeing all kinds of weird subdomain traffic including our staging server: staging.safarinow.com plus a bunch of other 'internal' ones.
Not really sure how these domains got out into the wild but they did, and we were receiving traffic from various bots including Googlebot for them.
So ... 301 redirected. Sorted.
Weird thing happened though.
Our overall server response time went down dramatically.
I dug into it via New Relic and found that our main accommodation search procedure throughput had more than halved.
My assumption is that each sub domain being hosted in its own appdomain, and each appdomain has its own EntLib cache
Once all traffic got consolidated to www, the cache suddenly came into its own.
Queries which once hit the database, now hit the cache. CPU on the database server is way down. Average server response has almost halved to around 2-300 milliseconds.
No comments:
Post a Comment