Current time: 12-13-2019, 08:31 AM Hello There, Guest! (LoginRegister)

Post Reply 
 
Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Can this result be improved?
02-05-2011, 01:18 AM
Post: #1
Can this result be improved?
http://www.webpagetest.org/result/110204...4c53f994e/

how do i improve this result?
Find all posts by this user
Quote this message in a reply
02-05-2011, 01:32 AM
Post: #2
RE: Can this result be improved?
I'd recommend improving the page first and then the result can be looked at. Even in a single request a little css styling and color would be nice.

That said, if you're serious about improving the performance of THAT page you can get a static IP, set the DNS TTL a lot higher (currently set at 1 minute) and use a dedicated DNS provider with a distributed anycast infrastructure.

You should also run a bunch of tests to make sure you don't get variability introduced by Dreamhost's servers. If the html is pretty static you can use a CDN and serve it directly from the edge.

-Pat
Visit this user's website Find all posts by this user
Quote this message in a reply
02-05-2011, 01:48 AM
Post: #3
RE: Can this result be improved?
That html only page? Maybe shorten the names of your folders so your htm-file becomes a bit smaller? :-)
Find all posts by this user
Quote this message in a reply
02-05-2011, 01:59 AM
Post: #4
RE: Can this result be improved?
Wooft! More of a shallow pond than a waterfall.

You might save a few milliseconds by reducing the size of your HTML. Chang your doctype to <!DOCTYPE HTML>. Very HTML5. You don't need quote marks for attributes so there's a couple of bytes.

http://www.bulletbits.com - My Performance Blog
Find all posts by this user
Quote this message in a reply
02-06-2011, 06:29 AM
Post: #5
RE: Can this result be improved?
Thanks for the points.

fyi, i want to optimize that SITE, not THAT page - but want to do so systematically.

an empty page with 1 request seems a logical place to start.

regardless, added some color to the page....
http://www.webpagetest.org/result/110205.../#request1

assuming the page is served by maxcdn, is the static ip address beneficial?

is the dedicated DNS provider necessary if the page is served by a cdn?

where it is showing ttl of 60? ping shows about 40s, dig says the 1074s for the A record and 14400 for the NS records.

re: reducing the size of the html by a few bytes
file size was 1.7KB, which downloaded in 0 ms.
Find all posts by this user
Quote this message in a reply
02-07-2011, 11:01 PM
Post: #6
RE: Can this result be improved?
is there a prefered dns provider? i found this, but don't know if it's better or worse than other companies out there which offer the same service:
http://www.loaddns.com/solutions/manageddnsbasic.aspx
Find all posts by this user
Quote this message in a reply
02-07-2011, 11:27 PM
Post: #7
RE: Can this result be improved?
I don't have a lot of experience with various DNS providers, just look for one that uses an Anycast network and has servers close to your customers (DYN and Ultra are the bigger players that I'm aware of, I'm using DNS Made Easy but mostly because they're cheap :-) ).

If your base HTML pages are also going to be served by a CDN then you don't have to worry about a DNS provider, just make sure the CDN provider you select has a good DNS architecture as well (scary as it is, the largest CDN provider doesn't use Anycast for their DNS so resolution performance can vary quite wildly).

A static IP could still help, even when using something like maxcdn (I believe they offer it as an option either through maxcdn or their netdna offerings) because it will eliminate one of the steps in the lookup (CNAME->A record).

That said, this is all on the REALLY extreme side of optimizing. If the site is going to be more than just static files with each page being a single request then there's going to be a lot more low-hanging-fruit than optimizing the DNS architecture - the DNS will only buy you a few ms on a single request (more if the existing DNS is not great and uncached like it was in the test you ran).

Here is the record I get using nslookup (d2 debug level):

Got answer (93 bytes):
HEADER:
opcode = QUERY, id = 25, rcode = NOERROR
header flags: response, want recursion, recursion avail.
questions = 1, answers = 0, authority records = 1, additional = 0

QUESTIONS:
http://www.jesseo.com, type = AAAA, class = IN
AUTHORITY RECORDS:
-> jesseo.com
type = SOA, class = IN, dlen = 49
ttl = 60 (1 min)
primary name server = ns1.dreamhost.com
responsible mail addr = hostmaster.dreamhost.com
serial = 2011020200
refresh = 16206 (4 hours 30 mins 6 secs)
retry = 1800 (30 mins)
expire = 1814400 (21 days)
default TTL = 14400 (4 hours)
Visit this user's website Find all posts by this user
Quote this message in a reply
02-09-2011, 12:45 AM
Post: #8
RE: Can this result be improved?
(02-07-2011 11:27 PM)pmeenan Wrote:  optimizing the DNS architecture will only buy you a few ms on a single request

ok - have set up dnsmadeeasy. thanks!
http://www.webpagetest.org/result/110208...1/details/

btw, i ran some tests and it seemed like the domain dnsmadeeasy.com was consistently about 0.2 seconds faster than the actual domain i'm optimizing. 55-85ms vs 220-370ms.

seems significant - as this is an absolute bottleneck.
Find all posts by this user
Quote this message in a reply
02-09-2011, 12:56 AM
Post: #9
RE: Can this result be improved?
DNS lookups for the dnsmadeeasy domain? Looked like the dns lookups for your domain in the test you linked were ~56ms.

Thanks,

-Pat
Visit this user's website Find all posts by this user
Quote this message in a reply
02-11-2011, 12:52 AM
Post: #10
RE: Can this result be improved?
Try out Amazons Route 53 . Been getting decent performance from them.

cdnplanet.com: We help you select the right CDN
Visit this user's website Find all posts by this user
Quote this message in a reply
Post Reply 


Forum Jump:


User(s) browsing this thread: 1 Guest(s)