Skip to search.

Breaking News Visit Yahoo! News for the latest.

×Close this window

exceptional-performance · Exceptional Performance
  • Members Only
  • Post
  • Links
  • Polls
  • Members
  • Calendar
  • Promote

The Yahoo! Groups Product Blog

Check it out!

Group Information

  • Members: 1380
  • Category: Yahoo!
  • Founded: Jul 18, 2007
  • Language: English
? Already a member? Sign in to Yahoo!

Yahoo! Groups Tips

Did you know...
Message search is now enhanced, find messages faster. Take it for a spin.

Messages
Message #
Search:
Advanced
 Messages Help
Messages
Topics
Hot Topics
Messages 1 - 30 of 2058   Oldest  |  < Older  |  Newer >  |  Newest
Messages: Show Message Summaries Sort by Date ^  
#1 From: "David" <dwash59@...>
Date: Wed Jul 25, 2007 3:10 am
Subject: error -207 bad package 		dwash59_2000
Send Email Send Email
  
Installing yslow from
https://addons.mozilla.org/en-US/firefox/downloads/file/17827/yslow-0.7.0-fx.xpi
doesn't work.

Reply | Messages in this Topic (3)
#2 From: "shiff2kl" <shiff2kl@...>
Date: Wed Jul 25, 2007 3:23 am
Subject: Re: error -207 bad package 		shiff2kl
Send Email Send Email
  
Download it from https://addons.mozilla.org/en-US/firefox/addon/5369

--- In exceptional-performance@yahoogroups.com, "David" <dwash59@...>
wrote:
>
> Installing yslow from
>
https://addons.mozilla.org/en-US/firefox/downloads/file/17827/yslow-0.7.0-fx.xpi
> doesn't work.
>

Reply | Messages in this Topic (3)
#3 From: "Steve Howard" <steve@...>
Date: Wed Jul 25, 2007 6:16 pm
Subject: Warning Message to Loosen Firefox Security Settings, How? 		howardsr
Send Email Send Email
  
When I run YSlow, it gives me a message at the top which says
"WARNING: For a full analysis you need to loosen Firefox's security
settings. Go to about:config and toggle
signed.applets.codebase_principal_support to true." I do not know
where to change these settings. The only "about" I know is Help ->
About which of course does not work. Also Tools -> Options does not
have anything like this. Please help. Looks like a nice tool.

Reply | Messages in this Topic (2)
#4 From: "swapnil.shinde" <sshinde@...>
Date: Wed Jul 25, 2007 7:29 pm
Subject: Re: Warning Message to Loosen Firefox Security Settings, How? 		swapnil.shinde
Send Email Send Email
 
Hi,

Please type "about:config" in the address bar of Firefox. This will open the Firefox configuration screen. Please type "signed.applets.codebase_principal_support" in the 'Filer' text box.

You should now see this configuration option. You can double-click this option to toggle it from 'false' to 'true'

Hope that helps.

Swapnil.

--- In exceptional-performance@yahoogroups.com, "Steve Howard" <steve@...> wrote:
>
> When I run YSlow, it gives me a message at the top which says
> "WARNING: For a full analysis you need to loosen Firefox's security
> settings. Go to about:config and toggle
> signed.applets.codebase_principal_support to true." I do not know
> where to change these settings. The only "about" I know is Help ->
> About which of course does not work. Also Tools -> Options does not
> have anything like this. Please help. Looks like a nice tool.
>

Reply | Messages in this Topic (2)
#5 From: "petebardo" <randy@...>
Date: Wed Jul 25, 2007 7:28 pm
Subject: Can't find YSlow 		petebardo
Send Email Send Email
  
I downloaded and installed Firebug, restarted Firefox, then downloaded
and installed YSlow. When I open Firebug, there's no performance tab
or any thing that looks like YSlow sample screens.

What'd I do wrong?

Reply | Messages in this Topic (4)
#6 From: "swapnil.shinde" <sshinde@...>
Date: Wed Jul 25, 2007 7:33 pm
Subject: Re: Can't find YSlow 		swapnil.shinde
Send Email Send Email
  
What is the Firefox version installed on your machine?

YSlow is compatible with FF 2.0+

Swapnil.

--- In exceptional-performance@yahoogroups.com, "petebardo" <randy@...>
wrote:
>
> I downloaded and installed Firebug, restarted Firefox, then downloaded
> and installed YSlow. When I open Firebug, there's no performance tab
> or any thing that looks like YSlow sample screens.
>
> What'd I do wrong?
>

Reply | Messages in this Topic (4)
#7 From: "swapnil.shinde" <sshinde@...>
Date: Wed Jul 25, 2007 7:31 pm
Subject: Re: error -207 bad package 		swapnil.shinde
Send Email Send Email
  
Hi David,

Thanks for reporting that issue. We have fixed it now. Sorry for the
inconvenience caused.

Swapnil.

--- In exceptional-performance@yahoogroups.com, "shiff2kl"
<shiff2kl@...> wrote:
>
> Download it from https://addons.mozilla.org/en-US/firefox/addon/5369
>
> --- In exceptional-performance@yahoogroups.com, "David" dwash59@
> wrote:
> >
> > Installing yslow from
> >
>
https://addons.mozilla.org/en-US/firefox/downloads/file/17827/yslow-0.7.\
0-fx.xpi
> > doesn't work.
> >
>

Reply | Messages in this Topic (3)
#8 From: "scottfrosty2000" <scottfrosty2000@...>
Date: Wed Jul 25, 2007 7:40 pm
Subject: Re: Can't find YSlow 		scottfrosty2000
Send Email Send Email
  
--- In exceptional-performance@yahoogroups.com, "petebardo" <randy@...>
wrote:
>
> I downloaded and installed Firebug, restarted Firefox, then downloaded
> and installed YSlow. When I open Firebug, there's no performance tab
> or any thing that looks like YSlow sample screens.
>
> What'd I do wrong?
>

Yslow is one of the tabs presented in firebug.  Select that tab and
then use the tabs above the firebug bar to review the stats.
Thanks,

-Scott

Reply | Messages in this Topic (4)
#9 From: "mxyzplkiv" <ernest.mueller@...>
Date: Wed Jul 25, 2007 8:09 pm
Subject: Glad it's out! 		mxyzplkiv
Send Email Send Email
  
I saw YSlow at Web 2.0 and we've been waiting eagerly for it to come
out (and the book to come out as well).  Thanks for releasing it to
the public!

Ernest Mueller
Web Systems Manager
National Instruments Corporation

Reply | Messages in this Topic (2)
#10 From: "swapnil.shinde" <sshinde@...>
Date: Wed Jul 25, 2007 8:53 pm
Subject: Re: Glad it's out! 		swapnil.shinde
Send Email Send Email
 
Glad to know you are enjoying YSlow :)

Swapnil.

--- In exceptional-performance@yahoogroups.com, "mxyzplkiv" <ernest.mueller@...> wrote:
>
> I saw YSlow at Web 2.0 and we've been waiting eagerly for it to come
> out (and the book to come out as well). Thanks for releasing it to
> the public!
>
> Ernest Mueller
> Web Systems Manager
> National Instruments Corporation
>

Reply | Messages in this Topic (2)
#11 From: "tully_tim" <timt@...>
Date: Wed Jul 25, 2007 10:00 pm
Subject: Script in head returning stale results? 		tully_tim
Send Email Send Email
  
Was playing with YSlow, and I moved some script's out of the head
section to an area definitely outside of it.  However, even after
clearing cache, enabling/disabling firebug, restarting firefox, yslow
is still reporting script in the head section. View source indicates
it's definitely outside of head.  Is YSlow still picking up stale
versions somehow?

Reply | Messages in this Topic (2)
#12 From: "petebardo" <randy@...>
Date: Wed Jul 25, 2007 10:38 pm
Subject: Re: Can't find YSlow 		petebardo
Send Email Send Email
  
It finally showed up after restarting Firefox a few times. Thanks for
the help.

Reply | Messages in this Topic (4)
#13 From: "bayareacoder" <bayareacoder@...>
Date: Thu Jul 26, 2007 12:36 am
Subject: YSlow config 		bayareacoder
Send Email Send Email
  
Great tool!  But since its already integrated into firebug can I take
it out of my status bar?  I'd rather just click the firebug icon (or
F12) and then click the YSlow tab.

Reply | Messages in this Topic (3)
#14 From: "swapnil.shinde" <sshinde@...>
Date: Thu Jul 26, 2007 12:57 am
Subject: Re: YSlow config 		swapnil.shinde
Send Email Send Email
  
We've gotten feedback that people don't want to see any of this info
in the statusbar. We'll add an option to hide the statusbar info in
the next release.

Thanks,
Swapnil

--- In exceptional-performance@yahoogroups.com, "bayareacoder"
<bayareacoder@...> wrote:
>
> Great tool!  But since its already integrated into firebug can I take
> it out of my status bar?  I'd rather just click the firebug icon (or
> F12) and then click the YSlow tab.
>

Reply | Messages in this Topic (3)
#15 From: "swapnil.shinde" <sshinde@...>
Date: Thu Jul 26, 2007 1:04 am
Subject: Re: Script in head returning stale results? 		swapnil.shinde
Send Email Send Email
  
YSlow will consider a script to be out of head section, if it is
anywhere after the <body> element in the document. Did you take the
script out of head and add it before <body>? If so, YSlow will still
flag this.

Swapnil.

--- In exceptional-performance@yahoogroups.com, "tully_tim" <timt@...>
wrote:
>
> Was playing with YSlow, and I moved some script's out of the head
> section to an area definitely outside of it.  However, even after
> clearing cache, enabling/disabling firebug, restarting firefox, yslow
> is still reporting script in the head section. View source indicates
> it's definitely outside of head.  Is YSlow still picking up stale
> versions somehow?
>

Reply | Messages in this Topic (2)
#16 From: "xertroyt" <xertroyt@...>
Date: Thu Jul 26, 2007 5:38 pm
Subject: Etags score real? 		xertroyt
Send Email Send Email
  
Hi.

I'm getting an "F" for ETags, but all of my images, css, .js, etc.,
have the tags . . .

What is the rule for the grade?

Reply | Messages in this Topic (8)
#17 From: Scott Frost <scottfrosty2000@...>
Date: Thu Jul 26, 2007 5:50 pm
Subject: Re: Etags score real? 		scottfrosty2000
Send Email Send Email
 
I'm not sure about the actual rule but I understand the logic behind the grade.  Click on the category itself and Yslow explains to you why setting ETag *could be* negative on overall performance.  Basically, ETags do keep content in the browser however if you have multiple servers the benefit diminishes.  I'd add to this explaination that ETags typically don't cache well because most cache servers see each object as unique.  It's important to use ETags with your overall, End-to-end web infrastructure in mind.
Thanks,
 
-Scott

----- Original Message ----
From: xertroyt <xertroyt@...>
To: exceptional-performance@yahoogroups.com
Sent: Thursday, July 26, 2007 10:38:38 AM
Subject: [exceptional-performance] Etags score real?

Hi.

I'm getting an "F" for ETags, but all of my images, css, .js, etc.,
have the tags . . .

What is the rule for the grade?



Boardwalk for $500? In 2007? Ha!
Play Monopoly Here and Now (it's updated for today's economy) at Yahoo! Games.

Reply | Messages in this Topic (8)
#18 From: Ernest Mueller <ernest.mueller@...>
Date: Thu Jul 26, 2007 6:01 pm
Subject: Re: Etags score real? 		mxyzplkiv
Send Email Send Email
  
Yeah, ETags are new to me and we're looking into it here at NI too...  It
appears that our Netscaler load balancer may be inserting ETags for us
which would make them "correct" despite multiple back end servers, I
believe.

Ernest
______________________
UN-altered REPRODUCTION and DISSEMINATION of
this IMPORTANT information is ENCOURAGED.




              Scott Frost
              <scottfrosty2000@
              yahoo.com>                                                 To
              Sent by:                  exceptional-performance@yahoogroups
              exceptional-perfo         .com
              rmance@yahoogroup                                          cc
              s.com
                                                                    Subject
                                        Re: [exceptional-performance] Etags
              07/26/2007 12:50          score real?
              PM


              Please respond to
              exceptional-perfo
              rmance@yahoogroup
                    s.com






I'm not sure about the actual rule but I understand the logic behind the
grade.  Click on the category itself and Yslow explains to you why setting
ETag *could be* negative on overall performance.  Basically, ETags do keep
content in the browser however if you have multiple servers the benefit
diminishes.  I'd add to this explaination that ETags typically don't cache
well because most cache servers see each object as unique.  It's important
to use ETags with your overall, End-to-end web infrastructure in mind.


Thanks,

-Scott

----- Original Message ----
From: xertroyt <xertroyt@...>
To: exceptional-performance@yahoogroups.com
Sent: Thursday, July 26, 2007 10:38:38 AM
Subject: [exceptional-performance] Etags score real?



Hi.

I'm getting an "F" for ETags, but all of my images, css, .js, etc.,
have the tags . . .

What is the rule for the grade?





Boardwalk for $500? In 2007? Ha!
Play Monopoly Here and Now (it's updated for today's economy) at Yahoo!
Games.

Reply | Messages in this Topic (8)
#19 From: "swapnil.shinde" <sshinde@...>
Date: Thu Jul 26, 2007 7:01 pm
Subject: Re: Etags score real? 		swapnil.shinde
Send Email Send Email
  
Etags should be removed. See here for more details:
http://developer.yahoo.com/performance/rules.html#etags

Thanks,
Swapnil.

--- In exceptional-performance@yahoogroups.com, "xertroyt"
<xertroyt@...> wrote:
>
> Hi.
>
> I'm getting an "F" for ETags, but all of my images, css, .js, etc.,
> have the tags . . .
>
> What is the rule for the grade?
>

Reply | Messages in this Topic (8)
#20 From: Ernest Mueller <ernest.mueller@...>
Date: Thu Jul 26, 2007 8:09 pm
Subject: Re: Re: Etags score real? 		mxyzplkiv
Send Email Send Email
  
Well, they should be removed if they are actually coming from individual
servers in a cluster, right?  If there's something (like a load balancer)
making the ETags "right" then they're of value I'd think - though I'm
unclear on the order of precedence of expires, cache-control, etag, etc
headers.  The "far future expires" thing sounds fine but in reality it's
hard to do thaton a large site...  If you could instead depend on an ETag
(aka checksum) you'd be at the happy medium, wouldn't you?

Ernest
______________________
UN-altered REPRODUCTION and DISSEMINATION of
this IMPORTANT information is ENCOURAGED.




              "swapnil.shinde"
              <sshinde@yahoo-in
              c.com>                                                     To
              Sent by:                  exceptional-performance@yahoogroups
              exceptional-perfo         .com
              rmance@yahoogroup                                          cc
              s.com
                                                                    Subject
                                        [exceptional-performance] Re: Etags
              07/26/2007 02:01          score real?
              PM


              Please respond to
              exceptional-perfo
              rmance@yahoogroup
                    s.com






Etags should be removed. See here for more details:
http://developer.yahoo.com/performance/rules.html#etags

Thanks,
Swapnil.

--- In exceptional-performance@yahoogroups.com, "xertroyt"
<xertroyt@...> wrote:
>
> Hi.
>
> I'm getting an "F" for ETags, but all of my images, css, .js, etc.,
> have the tags . . .
>
> What is the rule for the grade?
>

Reply | Messages in this Topic (8)
#21 From: Ernest Mueller <ernest.mueller@...>
Date: Thu Jul 26, 2007 8:59 pm
Subject: Our Performance Story 		mxyzplkiv
Send Email Send Email
  
In the interest of general Web site performance discussion, thought I'd
share some of our success stories on Web site performance!

We're not a super huge site of the Yahoo!, Amazon, etc. ilk, but we do put
through a couple million visits a month.  Our customers are heavily global
(the site's translated into 9 different languages, etc.) but the site is
hosted in the US.  The site has various technologies/application stacks -
static, Java, PL/SQL, Domino, etc.  We've had performance as an important
goal for a long time.

This year we made several changes that really helped our performance (see
the Keynote graph below - each line is a different technology stack on our
site).

(Embedded image moved to file: pic06851.jpg)
The first big improvement was when we moved from an older load balancer
software to new Citrix Netscalar load balancers.  This gave us compression,
caching, SSL offload, TCP multiplexing...  We saw great improvement across
the board, even on our purely static content (the light blue line).

The second big improvement was when we upgraded our technology stacks and
front end servers - we went from Sun V440/Solaris class to Dell 2950/Red
Hat class boxes on the front end after a performance bakeoff with some
other options (like the newer Sun AMD and Niagara servers).  Our big ol'
Java apps (the magenta line) saw huge improvement from that.

(And yes, we're the personal heroes of the business-side Web folks after
all this!  Yay IT!!!)

Now we're looking to work on the "front end" issues that the Yahoo! folks
have brought up (besides compression, which we got earlier).  We've known
for a long time that much of the time to user is based on "network effects"
- our rule of thumb is that one of our pages has a load time of n seconds
in the US, performance in Europe is n*2 and in Asia is n*3.  We knew we had
bigger fish to fry first, but now that those are done we're eager to see
what benefits we can get from those...  Like everyone else we're having
unprecedented growth in Asia and translated content doesn't help you if you
can't deliver it quickly!

I figured I'd jump in and share these metrics in the hope that others will
as well.  It's good to hear "performance tips" but it's hard to know
whether some tweak might get you a marginal improvement or a substantial
improvement, so I encourage people to share metrics on their changes for
everyone's benefit!

Thanks,

Ernest Mueller
Web Systems Manager
National Instruments Corporation
______________________
UN-altered REPRODUCTION and DISSEMINATION of
this IMPORTANT information is ENCOURAGED.

Reply | Messages in this Topic (1)
#22 From: Scott Frost <scottfrosty2000@...>
Date: Thu Jul 26, 2007 9:25 pm
Subject: Re: Re: Etags score real? 		scottfrosty2000
Send Email Send Email
 
Ernest,
Your NetScaler is setting ETags (which is basically a unique serial # for version of each object served) to the browser.  The value in doing this is that the browser saves fetching the actual object if the "serial" of the object in browser cache matches the "serial" of the current object being served from your Netscaler box. 
 
Since the ETag is unique to the Host header and the object name the Netscaler will serve a duplicate object for each origin server that sits behind it, (assuming the content is mirrored).  Assuming you update these objects often this could cause a substantial number of redundant objects in cache.
 
It also would hurt end user performance depending on how the Netscaler is performing the local load balacing, (IP versus cookie stickiness) and which server it hits. 
 
For example, let's say a given customer hits server A and pull object 1, 2 & 3 to the client browser.  All objects have a ETag of 3331 2 & 3 and they check for freshness every 5 minutes.  After 5 minutes your customer's browser checks freshness at origin for object 1. The Netscaler receives the request and routes it to server B.  Server B checks object 1's ETag and determines it's 4441.  Since 4441 doesn't match 3331 it determines the customer's broswer object is stale.  Thus, the browser calls again for the same object even though it has a current copy because the ETag didn't match.
 
This activity KILLs browser performance because these are all redundant calls.  This populates your Netscaler cache with lots of objects that only get hit once.  Thus your Netscaler works really hard, (assuming it's using LRU) to keep cached objects available and overall your cache hit miss ratio will be lower. 
 
The origin server farm will suffer from lots of unecessary requests which increases load and slows down first byte response and content download for your real traffic.
 
Also, If you use a CDN none of your objects will cache at the edge, you'll get more traffic to origin and all customer performance will suffer.
 
Does this help?
Thanks,
 
-Scott
 
 
 
----- Original Message ----
From: Ernest Mueller <ernest.mueller@...>
To: exceptional-performance@yahoogroups.com
Cc: exceptional-performance@yahoogroups.com
Sent: Thursday, July 26, 2007 1:09:21 PM
Subject: Re: [exceptional-performance] Re: Etags score real?


Well, they should be removed if they are actually coming from individual
servers in a cluster, right? If there's something (like a load balancer)
making the ETags "right" then they're of value I'd think - though I'm
unclear on the order of precedence of expires, cache-control, etag, etc
headers. The "far future expires" thing sounds fine but in reality it's
hard to do thaton a large site... If you could instead depend on an ETag
(aka checksum) you'd be at the happy medium, wouldn't you?

Ernest
____________ _________ _
UN-altered REPRODUCTION and DISSEMINATION of
this IMPORTANT information is ENCOURAGED.

"swapnil.shinde"
<sshinde@yahoo- in
c.com> To
Sent by: exceptional- performance@ yahoogroups
exceptional- perfo .com
rmance@yahoogroup cc
s.com
Subject
[exceptional- performance] Re: Etags
07/26/2007 02:01 score real?
PM


Please respond to
exceptional- perfo
rmance@yahoogroup
s.com



Etags should be removed. See here for more details:
http://developer. yahoo.com/ performance/ rules.html# etags

Thanks,
Swapnil.

--- In exceptional- performance@ yahoogroups. com, "xertroyt"
<xertroyt@.. .> wrote:
>
> Hi.
>
> I'm getting an "F" for ETags, but all of my images, css, .js, etc.,
> have the tags . . .
>
> What is the rule for the grade?
>



Be a better Globetrotter. Get better travel answers from someone who knows.
Yahoo! Answers - Check it out.

Reply | Messages in this Topic (8)
#23 From: Ernest Mueller <ernest.mueller@...>
Date: Thu Jul 26, 2007 9:54 pm
Subject: Re: Re: Etags score real? 		mxyzplkiv
Send Email Send Email
  
Gotcha.  I wasn't sure whether the Netscaler was being "smart" and deduping
etags per cluster or not.  I'm going to confirm with support.  If it or a
similar technology could (F5 etc) it would then be good, right?

Ernest
______________________
UN-altered REPRODUCTION and DISSEMINATION of
this IMPORTANT information is ENCOURAGED.




              Scott Frost
              <scottfrosty2000@
              yahoo.com>                                                 To
              Sent by:                  exceptional-performance@yahoogroups
              exceptional-perfo         .com
              rmance@yahoogroup                                          cc
              s.com
                                                                    Subject
                                        Re: [exceptional-performance] Re:
              07/26/2007 04:25          Etags score real?
              PM


              Please respond to
              exceptional-perfo
              rmance@yahoogroup
                    s.com






Ernest,


Your NetScaler is setting ETags (which is basically a unique serial # for
version of each object served) to the browser.  The value in doing this is
that the browser saves fetching the actual object if the "serial" of the
object in browser cache matches the "serial" of the current object being
served from your Netscaler box.

Since the ETag is unique to the Host header and the object name the
Netscaler will serve a duplicate object for each origin server that sits
behind it, (assuming the content is mirrored).  Assuming you update these
objects often this could cause a substantial number of redundant objects in
cache.

It also would hurt end user performance depending on how the Netscaler is
performing the local load balacing, (IP versus cookie stickiness) and which
server it hits.

For example, let's say a given customer hits server A and pull object 1, 2
& 3 to the client browser.  All objects have a ETag of 3331 2 & 3 and they
check for freshness every 5 minutes.  After 5 minutes your customer's
browser checks freshness at origin for object 1. The Netscaler receives the
request and routes it to server B.  Server B checks object 1's ETag and
determines it's 4441.  Since 4441 doesn't match 3331 it determines the
customer's broswer object is stale.  Thus, the browser calls again for the
same object even though it has a current copy because the ETag didn't
match.

This activity KILLs browser performance because these are all redundant
calls.  This populates your Netscaler cache with lots of objects that only
get hit once.  Thus your Netscaler works really hard, (assuming it's using
LRU) to keep cached objects available and overall your cache hit miss ratio
will be lower.

The origin server farm will suffer from lots of unecessary requests which
increases load and slows down first byte response and content download for
your real traffic.

Also, If you use a CDN none of your objects will cache at the edge, you'll
get more traffic to origin and all customer performance will suffer.

Does this help?
Thanks,

-Scott



----- Original Message ----
From: Ernest Mueller <ernest.mueller@...>
To: exceptional-performance@yahoogroups.com
Cc: exceptional-performance@yahoogroups.com
Sent: Thursday, July 26, 2007 1:09:21 PM
Subject: Re: [exceptional-performance] Re: Etags score real?




Well, they should be removed if they are actually coming from individual
servers in a cluster, right? If there's something (like a load balancer)
making the ETags "right" then they're of value I'd think - though I'm
unclear on the order of precedence of expires, cache-control, etag, etc
headers. The "far future expires" thing sounds fine but in reality it's
hard to do thaton a large site... If you could instead depend on an ETag
(aka checksum) you'd be at the happy medium, wouldn't you?

Ernest
____________ _________ _
UN-altered REPRODUCTION and DISSEMINATION of
this IMPORTANT information is ENCOURAGED.

"swapnil.shinde"
<sshinde@yahoo- in
c.com> To
Sent by: exceptional- performance@ yahoogroups
exceptional- perfo .com
rmance@yahoogroup cc
s.com
Subject
[exceptional- performance] Re: Etags
07/26/2007 02:01 score real?
PM


Please respond to
exceptional- perfo
rmance@yahoogroup
s.com



Etags should be removed. See here for more details:
http://developer. yahoo.com/ performance/ rules.html# etags

Thanks,
Swapnil.

--- In exceptional- performance@ yahoogroups. com, "xertroyt"
<xertroyt@.. .> wrote:
>
> Hi.
>
> I'm getting an "F" for ETags, but all of my images, css, .js, etc.,
> have the tags . . .
>
> What is the rule for the grade?
>





Be a better Globetrotter. Get better travel answers from someone who knows.
Yahoo! Answers - Check it out.

Reply | Messages in this Topic (8)
#24 From: Scott Frost <scottfrosty2000@...>
Date: Fri Jul 27, 2007 12:41 am
Subject: Re: Re: Etags score real? 		scottfrosty2000
Send Email Send Email
 
In my experience this isn't something a LB box can solve.  People use IP or cookie stickiness to tie a session to a particular origin web server. 
 
The Netscaler would have to ignore the ETag heading or re-write the ETag to do what your suggesting.  This overhead would make ETags pointless and suggests to me to use cache control If-Modified-Since instead. 


----- Original Message ----
From: Ernest Mueller <ernest.mueller@...>
To: exceptional-performance@yahoogroups.com
Cc: exceptional-performance@yahoogroups.com
Sent: Thursday, July 26, 2007 2:54:43 PM
Subject: Re: [exceptional-performance] Re: Etags score real?


Gotcha. I wasn't sure whether the Netscaler was being "smart" and deduping
etags per cluster or not. I'm going to confirm with support. If it or a
similar technology could (F5 etc) it would then be good, right?

Ernest
____________ _________ _
UN-altered REPRODUCTION and DISSEMINATION of
this IMPORTANT information is ENCOURAGED.

Scott Frost
<scottfrosty2000@
yahoo.com> To
Sent by: exceptional- performance@ yahoogroups
exceptional- perfo .com
rmance@yahoogroup cc
s.com
Subject
Re: [exceptional- performance] Re:
07/26/2007 04:25 Etags score real?
PM


Please respond to
exceptional- perfo
rmance@yahoogroup
s.com



Ernest,

Your NetScaler is setting ETags (which is basically a unique serial # for
version of each object served) to the browser. The value in doing this is
that the browser saves fetching the actual object if the "serial" of the
object in browser cache matches the "serial" of the current object being
served from your Netscaler box.

Since the ETag is unique to the Host header and the object name the
Netscaler will serve a duplicate object for each origin server that sits
behind it, (assuming the content is mirrored). Assuming you update these
objects often this could cause a substantial number of redundant objects in
cache.

It also would hurt end user performance depending on how the Netscaler is
performing the local load balacing, (IP versus cookie stickiness) and which
server it hits.

For example, let's say a given customer hits server A and pull object 1, 2
& 3 to the client browser. All objects have a ETag of 3331 2 & 3 and they
check for freshness every 5 minutes. After 5 minutes your customer's
browser checks freshness at origin for object 1. The Netscaler receives the
request and routes it to server B. Server B checks object 1's ETag and
determines it's 4441. Since 4441 doesn't match 3331 it determines the
customer's broswer object is stale. Thus, the browser calls again for the
same object even though it has a current copy because the ETag didn't
match.

This activity KILLs browser performance because these are all redundant
calls. This populates your Netscaler cache with lots of objects that only
get hit once. Thus your Netscaler works really hard, (assuming it's using
LRU) to keep cached objects available and overall your cache hit miss ratio
will be lower.

The origin server farm will suffer from lots of unecessary requests which
increases load and slows down first byte response and content download for
your real traffic.

Also, If you use a CDN none of your objects will cache at the edge, you'll
get more traffic to origin and all customer performance will suffer.

Does this help?
Thanks,

-Scott

----- Original Message ----
From: Ernest Mueller <ernest.mueller@ ni.com>
To: exceptional- performance@ yahoogroups. com
Cc: exceptional- performance@ yahoogroups. com
Sent: Thursday, July 26, 2007 1:09:21 PM
Subject: Re: [exceptional- performance] Re: Etags score real?

Well, they should be removed if they are actually coming from individual
servers in a cluster, right? If there's something (like a load balancer)
making the ETags "right" then they're of value I'd think - though I'm
unclear on the order of precedence of expires, cache-control, etag, etc
headers. The "far future expires" thing sounds fine but in reality it's
hard to do thaton a large site... If you could instead depend on an ETag
(aka checksum) you'd be at the happy medium, wouldn't you?

Ernest
____________ _________ _
UN-altered REPRODUCTION and DISSEMINATION of
this IMPORTANT information is ENCOURAGED.

"swapnil.shinde"
<sshinde@yahoo- in
c.com> To
Sent by: exceptional- performance@ yahoogroups
exceptional- perfo .com
rmance@yahoogroup cc
s.com
Subject
[exceptional- performance] Re: Etags
07/26/2007 02:01 score real?
PM

Please respond to
exceptional- perfo
rmance@yahoogroup
s.com

Etags should be removed. See here for more details:
http://developer. yahoo.com/ performance/ rules.html# etags

Thanks,
Swapnil.

--- In exceptional- performance@ yahoogroups. com, "xertroyt"
<xertroyt@.. .> wrote:
>
> Hi.
>
> I'm getting an "F" for ETags, but all of my images, css, .js, etc.,
> have the tags . . .
>
> What is the rule for the grade?
>

Be a better Globetrotter. Get better travel answers from someone who knows.
Yahoo! Answers - Check it out.



Looking for a deal? Find great prices on flights and hotels with Yahoo! FareChase.

Reply | Messages in this Topic (8)
#25 From: "pabloimpallari" <pabloimpallari@...>
Date: Fri Jul 27, 2007 11:11 am
Subject: Re: YSlow config 		pabloimpallari
Send Email Send Email
  
I want it out of the status bar too!
And if the tools option popups new tabs instead of new windows will be
grat too!

--- In exceptional-performance@yahoogroups.com, "bayareacoder"
<bayareacoder@...> wrote:
>
> Great tool!  But since its already integrated into firebug can I take
> it out of my status bar?  I'd rather just click the firebug icon (or
> F12) and then click the YSlow tab.
>

Reply | Messages in this Topic (3)
#26 From: "dglead" <dglead@...>
Date: Thu Aug 2, 2007 12:00 am
Subject: Bug on resources with no Content-Length in YSlow 		dglead
Send Email Send Email
  
If a page is serving a JavaScript file dynamically and not including a
Content-Length header it seems that YSlow takes it as having a size of
0 bytes.

You can see this on http://hiveminder.com/ (that site is serving a
JavaScript file, 220KB gzipped, 1068KB uncompressed).

Reply | Messages in this Topic (2)
#27 From: "Peter Bengtsson" <mail@...>
Date: Mon Aug 6, 2007 9:32 am
Subject: Expires header (and Cache-Control instead) 		peter_bengts...
Send Email Send Email
  
First of all a great tool guys! Thanks.

YSlow complains about objects that "do not have a far future Expires
header". Eg. our brochure site at www.fry-it.com

First of all, Cache-Control is a more modern way of setting
"cache-till-future" settings since you don't have to specify a date
format with all its risks of getting timezones, RFC formats, etc wrong.

Secondly, I added Expires headers too now and even though they're
several hours ahead of now YSlow still complains. I don't dare to make
it weeks and weeks into the future since if I need to change them
that'll cause problems for a long time till the old objects die.

PS. I've started coding on a solution that automatically calls the JS
and CSS by a version number in the filename so that I can set it to
expire 2010.

Reply | Messages in this Topic (2)
#28 From: "Nick Le Mouton" <noodles@...>
Date: Wed Aug 8, 2007 1:42 am
Subject: gzip js and css 		southofheaven
Send Email Send Email
  
Hi,

Rule 4 of the 13 performance rules talks about gzip'ing components, but
there are a few browsers that have problems with gzip'ed js/css.

I did some searching on the internet and found the following for Apache:

# Netscape 4.x or IE 5.5/6.0
BrowserMatch ^Mozilla/4 no-gzip
# IE 5.5 and IE 6.0 have bugs! Ignore them until IE 7.0+
BrowserMatch \bMSIE\s7 !no-gzip
# IE 6.0 after SP2 has no gzip bugs!
BrowserMatch \bMSIE.*SV !no-gzip
# Sometimes Opera pretends to be IE with "Mozila/4.0"
BrowserMatch \bOpera !no-gzip
AddOutputFilterByType DEFLATE text/css
AddOutputFilterByType DEFLATE text/javascript
AddOutputFilterByType DEFLATE application/x-javascript
Header append Vary User-Agent

Does this cover all the browsers that have problems or are there some that
are missing?

Is there also a lighttpd version of this (as I'm looking at moving our
images/static files to a separate lighttpd server).

Thanks
Nick

Reply | Messages in this Topic (6)
#29 From: "scottfrosty2000" <scottfrosty2000@...>
Date: Wed Aug 8, 2007 6:38 pm
Subject: Re: gzip js and css 		scottfrosty2000
Send Email Send Email
  
--- In exceptional-performance@yahoogroups.com, "Nick Le Mouton"
<noodles@...> wrote:
>
> Hi,
>
> Rule 4 of the 13 performance rules talks about gzip'ing components,
but
> there are a few browsers that have problems with gzip'ed js/css.
>
> I did some searching on the internet and found the following for
Apache:
>
> # Netscape 4.x or IE 5.5/6.0
> BrowserMatch ^Mozilla/4 no-gzip
> # IE 5.5 and IE 6.0 have bugs! Ignore them until IE 7.0+
> BrowserMatch \bMSIE\s7 !no-gzip
> # IE 6.0 after SP2 has no gzip bugs!
> BrowserMatch \bMSIE.*SV !no-gzip
> # Sometimes Opera pretends to be IE with "Mozila/4.0"
> BrowserMatch \bOpera !no-gzip
> AddOutputFilterByType DEFLATE text/css
> AddOutputFilterByType DEFLATE text/javascript
> AddOutputFilterByType DEFLATE application/x-javascript
> Header append Vary User-Agent
>
> Does this cover all the browsers that have problems or are there
some that
> are missing?
>
> Is there also a lighttpd version of this (as I'm looking at moving
our
> images/static files to a separate lighttpd server).
>
> Thanks
> Nick
>


Nick,
I think it's better to address your question by looking at what most
users are browsing with.

IMHO, the benefit that a user gets from sending compressed images to
the browser is greater than the problems it causes.  According to W3c
the 57% of users are on IE6.0 or higher.  When patched properly they
don't encounter issues with compressed content.  Apache can use vary
headers to work around browsers that don't support it.  Firefox (34%
of the browser share)supports HTML compression and images too (pretty
sure).  These two browser represent over 90% of total user desktops.


I'd suggest being careful about compressing .js and .css files as
this is where most of the problems lie.  Images and HTML are pretty
safe and provide significant performance benefit.


As for your question about a separate agent for YSlow:

I'm not sure one would want a separate version of YSlow for just for
lighttpd?  Lighttpd is a light weight webserver designed to serve
static content quickly.  You call the images, etc from the Lighttpd
domain from your main site because it's more efficent than serving
them off a full featured web server.

If we use YSlow to capture main site performance we'd see Lighttpd's
benefit through better main site performance.  This makes sense
because it's how a user hitting your site would see it.

If I wanted to measure the performance of Lighttpd directly I'd
exclude any non-Lighttpd domain URL's from the test.  I don't think
there is a domain/URL exclusion feature in YSlow.  I suppose you
could write  a page with only Lighttpd images and measure it but it
would defeat the point.

In the end, I believe one needs to measure the main site as a whole.
Lighttpd is just supporting infrastructure to make your end user
experience fast.  This should come through in the overall site
performance measurement.
Best Regards,

-Scott

Reply | Messages in this Topic (6)
#30 From: "Nick Le Mouton" <noodles@...>
Date: Wed Aug 8, 2007 10:08 pm
Subject: RE: Re: gzip js and css 		southofheaven
Send Email Send Email
  
Hi Scott,

Compressing images using gzip has been noted throughout the internet as a
waste of CPU cycles on both ends as a lot of the gif, jpg and png don't
compress much at all. I never brought up gzip'ing images for this reason.

Unfortunately with the site that I'm working on we can't ignore 10% of the
users, hoping that the js/css turns out ok for them. I wanted to know if
there were any main stream browsers that had problems with compressed js/css
that weren't listed.

The browsers I listed for no gzip on js/css were:
Netscape 4.x or IE 5.5/6.0

Apparently the following are ok with gzip on js/css:
IE 6.0 SP2+
IE 7
Firefox
Opera

I wasn't asking for another version of YSlow, I was asking for a config that
would do the same thing in lighttpd as the config changes I listed for
Apache.

Nick

________________________________________
From: exceptional-performance@yahoogroups.com
[mailto:exceptional-performance@yahoogroups.com] On Behalf Of
scottfrosty2000
Sent: Thursday, 9 August 2007 6:38 a.m.
To: exceptional-performance@yahoogroups.com
Subject: [exceptional-performance] Re: gzip js and css

Nick,
I think it's better to address your question by looking at what most
users are browsing with.

IMHO, the benefit that a user gets from sending compressed images to
the browser is greater than the problems it causes. According to W3c
the 57% of users are on IE6.0 or higher. When patched properly they
don't encounter issues with compressed content. Apache can use vary
headers to work around browsers that don't support it. Firefox (34%
of the browser share)supports HTML compression and images too (pretty
sure). These two browser represent over 90% of total user desktops.

I'd suggest being careful about compressing .js and .css files as
this is where most of the problems lie. Images and HTML are pretty
safe and provide significant performance benefit.

As for your question about a separate agent for YSlow:

I'm not sure one would want a separate version of YSlow for just for
lighttpd? Lighttpd is a light weight webserver designed to serve
static content quickly. You call the images, etc from the Lighttpd
domain from your main site because it's more efficent than serving
them off a full featured web server.

If we use YSlow to capture main site performance we'd see Lighttpd's
benefit through better main site performance. This makes sense
because it's how a user hitting your site would see it.

If I wanted to measure the performance of Lighttpd directly I'd
exclude any non-Lighttpd domain URL's from the test. I don't think
there is a domain/URL exclusion feature in YSlow. I suppose you
could write a page with only Lighttpd images and measure it but it
would defeat the point.

In the end, I believe one needs to measure the main site as a whole.
Lighttpd is just supporting infrastructure to make your end user
experience fast. This should come through in the overall site
performance measurement.
Best Regards,

-Scott

Reply | Messages in this Topic (6)
Messages 1 - 30 of 2058   Oldest  |  < Older  |  Newer >  |  Newest
Add to My Yahoo!      XML What's This?
Copyright © 2010 Yahoo! Inc. All rights reserved.
Privacy Policy - Terms of Service - Guidelines NEW - Help