This is the core business model for Keynote and Gomez.  Many CDNs will
arrange for a temp account with one of these to validate the benefit
they'll bring you during the sales cycle.  (Hint:  tell them to.)

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




              "Nick Le Mouton"
              <noodles@planetsl
              ackers.com>                                                To
              Sent by:                  <exceptional-performance@yahoogroup
              exceptional-perfo         s.com>
              rmance@yahoogroup                                          cc
              s.com
                                                                    Subject
                                        [exceptional-performance] Latency
              03/31/2008 05:37          measurements
              PM


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






Does anyone know of a service which allows you to check your web site's
connectivity/latency from multiple locations? We have our site hosted in a
single datacenter on the east coast and are considering using a CDN but
need
to know how much of a speed up we're likely to get if we move the content
closer to our end users (the majority of which are US based).

Thanks
Nick

Any chance we can get a version of Yslow that works with beta5. The
0.9.4 version is incompatible and cannot be (easily) installed.

Hi all,

A few months ago I was playing round with a JS/Ajax/php data-table (as
a faster solution to XML Databinding)...  (example
http://www.icommunicate.co.uk/tables/)

We stumbled upon a big problem with getting IE to behave and cache the
AJAX requests that are being made in the background - basically we
couldn't the eTags hooked up..

Anyway - having watched Steve Souders great talk on YouTube and read
the bit about eTag escaping in
http://developer.yahoo.com/performance/rules.html#etags ("The only
format constraints are that the string be quoted.") everything seems
to be working in IE as well as it was in FF! :-)

Just wanted to say thanks and, to anyone else that's trying to use
eTags to control whether to 304 AJAX requests that little note is
pretty important!

HTH,
Jimbo

There was a thread about this a while ago, started by me following
seeing a blog post suggesting that disabling Last-Modified would help,
as browsers would be more inclined to honour the Expires (or even
Etags) - which I think is what you are suggesting.

However on testing (bothing a developerment site with IE and Firefox,
and on a live website - with real traffic) it was not really the case.
Even with Expires set browsers would request the object again more
often at least when the user pressed F5 - possibly other cases (I
observe a high rate of them, but I could only reproduce pressing F5) -
or it could be users disabled caching (IE offers this )*

So my conclusion was to leave them on - at least that way the
webserver can usually return a 304 for these repeat requests.

Yes I know Etags are meant to offer similar functionality, but from
what I can tell they are not as widly supported as Last-Modified.

One thing I learnt is the Last-Modifed check at the server end can be
a simple string comparison, I was parsing the date and comparing
numbers, but that is not needed, the browser seems to store it as a
string. This also allows you to use fake dates, if you dont want to
disclose real dates (if that is your objection to Last-Modified) -
although someone please correct me if this is not good idea!


* One mitigating circumstance which only came to light later, is the
static content was served from a subdomain of the main domain, so
cookies where still used (we didnt define a server name) - so that
could of lowered the browsers reliance on Expires. Will have to test
it again nwo no cookies are invoved in static content.




On Tue, Apr 8, 2008 at 12:11 PM, Rakesh Rajan <rakeshxp@...> wrote:
>
>  We are trying to implement some of the performance rules in LifeBlob.
>
> A sample header ( for a resource file ) looks like :
>
>
> DateTue, 08 Apr 2008 11:08:16 GMT
> ServerApache
> Last-ModifiedWed, 02 Apr 2008 19:15:29 GMT
> Accept-Ranges bytes
> Cache-Controlmax-age=34560000
> ExpiresWed, 08 Apr 2009 11:08:16 GMT
> VaryAccept-Encoding
> Content-Encodinggzip
>  Content-Length9236
> Connectionclose
>  Content-Typetext/css
> So the question I have is around Last-Modified header. Is this header really
> necessary since I have all the caching headers present?  Also, I am planning
> to turn on ETags.
>
> Thanks,
> Rakesh Rajan
>



--
Barry

- www.nearby.org.uk - www.geograph.org.uk -

Rakesh Rajan wrote:
> I missed a point in the last mail. When I looked at the headers in case of
> Yahoo, they too include Last-Modified ( inspite of having all caching
> headers). So what is the advantage of having this header ? ( Assuming ETag
> is on )

Last-Modified and ETag headers are quite different from cache control
headers. They are used to _Revalidate_ cached copies of objects in a
conditional GET (note that it's quite likely that an expired copy is
still a valid copy).

You might want to enable both Last-Modified and ETag in anticipation of
cases when it's possible for an object to be "modified" but not actually
changed or significantly enough to be considered changed, for example:
on unix you can touch a file to change its modified timestamp without
altering the content.


> -Rakesh


--
Zhang Yining
URL:         http://www.zhangyining.net | http://www.yining.org
mailto:      yining@... | zhang.yining@...
Fingerprint: 25C8 47AE 30D5 4C0D A4BB  8CF2 3C2D 585F A905 F033

On Wed, Apr 9, 2008 at 11:38 AM, Rakesh Rajan <rakeshxp@...> wrote:
>
>  Zhang ,
>
>
> >
> >
> >
> >
> >
> >
> > You might want to enable both Last-Modified and ETag in anticipation of
> > cases when it's possible for an object to be "modified" but not actually
> > changed or significantly enough to be considered changed, for example:
> > on unix you can touch a file to change its modified timestamp without
> > altering the content.
> >
> >
> >
> >
> >
> >
> >
> > Recent Activity
> >
> >
> >  20
> > New Members Visit Your Group
> >
> >
> > Yahoo! Groups
> >
> >
>
>  Does this mean that even if only the timestamp of the file is changed ( and
> the content is still the same ), the server would return a  304 ( even
> though last-modified header is changed ) ?
>

Do you mean that Etags will do that? Well it depends, properly
configured etags, that just depend on the content, would do that.

But from what I gather (and I know Apache does) the default
configuration includes the timestamp in the Etag generation. (as it
uses the inode modification date I believe) - but its configurable.
This is one of the reasons YSlow recommends disabling etags as they
can be harder to get right, and often just add bloat (as they
essentially duplicate the Last-Modified)

HTH,

Barry


> -Rakesh
> >
> >
> >
> > D. __,_._,__
>
>



--
Barry

- www.nearby.org.uk - www.geograph.org.uk -

Rakesh Rajan wrote:
>  Does this mean that even if only the timestamp of the file is changed ( and
> the content is still the same ), the server would return a  304 ( even
> though last-modified header is changed ) ?

IIRC, if a request comes with both Last-Modified-Since and
If-None-Match, then the server should check both.

So for such requests, if the modified time has changed for the requested
resource, then the cache is considered invalid, and the whole entity
body will be returned instead of a 304, without even checking the etag.

So if you want to re-validate the cache with etag only, then make the
server return the etag header only. In that case, you should check how
the etag is generated on your server (on Apache, the modified time is
part of the etag value generation by default).


--
Zhang Yining
URL:         http://www.zhangyining.net | http://www.yining.org
mailto:      yining@... | zhang.yining@...
Fingerprint: 25C8 47AE 30D5 4C0D A4BB  8CF2 3C2D 585F A905 F033

On Wed, Apr 9, 2008 at 2:40 PM, Brian Williams <brian@...> wrote:
>
>  That would go back to an American grading system for schools
>
> From best to worst it goes from A to F skipping E for some reason.. not to
> sure on why the system does that.
>

<OT>
In the UK, its A-D* then F(ail) then U(nclassified)

(although sometimes A-E is used)
</OT>

>
>
>
>
> On Wed, Apr 9, 2008 at 8:39 AM, Reham El Gammal <reham.elgammal@...>
> wrote:
>
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > Hello
> >
> > What is the meaning of Yslow grades "A", "B" and "C"
> >
> >
> >
> > Thanks
> >
> > Reham Elgammal
> >
> >
>
>



--
Barry

- www.nearby.org.uk - www.geograph.org.uk -

Rakesh Rajan wrote:
> Does this mean that even if only the timestamp of the file is changed ( and
> the content is still the same ), the server would return a 304 ( even
> though last-modified header is changed ) ?

IIRC, if a request comes with both Last-Modified-Since and
If-None-Match, then the server should check both.

So for such requests, if the modified time has changed for the requested
resource, then the cache is considered invalid, and the whole entity
body will be returned instead of a 304, without even checking the etag.

So if you want to re-validate the cache with etag only, then make the
server return the etag header only. In that case, you should check how
the etag is generated on your server (on Apache, the modified time is
part of the etag value generation by default).

--
Zhang Yining
URL: http://www.zhangyining.net | http://www.yining.org
mailto: yining@zhangyining.net | zhang.yining@gmail.com
Fingerprint: 25C8 47AE 30D5 4C0D A4BB 8CF2 3C2D 585F A905 F033

Steve Souders wrote:
> ETag takes precedence over Last-Modified. In other words, if the
> If-None-Match test fails then the server doesn't even do the
> If-Modified-Since check.

You are right, and looking into apache source code[0] also confirms
this. Thanks for correcting my mistake.

However, what we are discussing here is behavior of implementation(s), I
have not looked into the RFC yet, but here is the quote from the book
"RESTful Web Services" (para 2, p247):

"If a server provides both Last-Modified and Etag, the client can
provide both If-Modified-Since and If-None-Match in subsequent requests
(as I did in Example 8-4). The server should make both checks: it should
only send a new representation if the representation has changed and the
Etag is different."

I believe this is worth pointing out, although it's probably outside the
scope of this group as it's mainly meant for handling dynamically
generated 'resources'.

- Yining

[0] http://www.srcdoc.com/apache_2.0.54/http__protocol_8c-source.html

> But in most cases this is moot because, as has
> been pointed out, the default ETag configuration contains the
> last-modified date.
>
> -Steve
>
>
> On 4/9/2008 5:20 AM, Zhang Yining wrote:
>>
>> Rakesh Rajan wrote:
>> > Does this mean that even if only the timestamp of the file is
>> changed ( and
>> > the content is still the same ), the server would return a 304 ( even
>> > though last-modified header is changed ) ?
>>
>> IIRC, if a request comes with both Last-Modified-Since and
>> If-None-Match, then the server should check both.
>>
>> So for such requests, if the modified time has changed for the requested
>> resource, then the cache is considered invalid, and the whole entity
>> body will be returned instead of a 304, without even checking the etag.
>>
>> So if you want to re-validate the cache with etag only, then make the
>> server return the etag header only. In that case, you should check how
>> the etag is generated on your server (on Apache, the modified time is
>> part of the etag value generation by default).
>>
>> --
>> Zhang Yining
>> URL: http://www.zhangyining.net <http://www.zhangyining.net> |
>> http://www.yining.org <http://www.yining.org>
>> mailto: yining@... <mailto:yining%40zhangyining.net> |
>> zhang.yining@... <mailto:zhang.yining%40gmail.com>
>> Fingerprint: 25C8 47AE 30D5 4C0D A4BB 8CF2 3C2D 585F A905 F033
>>
>>
>


--
Zhang Yining
URL:         http://www.zhangyining.net | http://www.yining.org
mailto:      yining@... | zhang.yining@...
Fingerprint: 25C8 47AE 30D5 4C0D A4BB  8CF2 3C2D 585F A905 F033

Steve Souders wrote:
> ETag takes precedence over Last-Modified. In other words, if the
> If-None-Match test fails then the server doesn't even do the
> If-Modified-Since check.

You are right, and looking into apache source code[0] also confirms
this. Thanks for correcting my mistake.

However, what we are discussing here is behavior of implementation(s), I
have not looked into the RFC yet, but here is the quote from the book
"RESTful Web Services" (para 2, p247):

"If a server provides both Last-Modified and Etag, the client can
provide both If-Modified-Since and If-None-Match in subsequent requests
(as I did in Example 8-4). The server should make both checks: it should
only send a new representation if the representation has changed and the
Etag is different."

I believe this is worth pointing out, although it's probably outside the
scope of this group as it's mainly meant for handling dynamically
generated 'resources'.

- Yining

[0] http://www.srcdoc.com/apache_2.0.54/http__protocol_8c-source.html

> But in most cases this is moot because, as has
> been pointed out, the default ETag configuration contains the
> last-modified date.
>
> -Steve
>
>
> On 4/9/2008 5:20 AM, Zhang Yining wrote:
>>
>> Rakesh Rajan wrote:
>> > Does this mean that even if only the timestamp of the file is
>> changed ( and
>> > the content is still the same ), the server would return a 304 ( even
>> > though last-modified header is changed ) ?
>>
>> IIRC, if a request comes with both Last-Modified-Since and
>> If-None-Match, then the server should check both.
>>
>> So for such requests, if the modified time has changed for the requested
>> resource, then the cache is considered invalid, and the whole entity
>> body will be returned instead of a 304, without even checking the etag.
>>
>> So if you want to re-validate the cache with etag only, then make the
>> server return the etag header only. In that case, you should check how
>> the etag is generated on your server (on Apache, the modified time is
>> part of the etag value generation by default).
>>
>> --
>> Zhang Yining
>> URL: http://www.zhangyining.net <http://www.zhangyining.net> |
>> http://www.yining.org <http://www.yining.org>
>> mailto: yining@zhangyining.net <mailto:yining%40zhangyining.net> |
>> zhang.yining@gmail.com <mailto:zhang.yining%40gmail.com>
>> Fingerprint: 25C8 47AE 30D5 4C0D A4BB 8CF2 3C2D 585F A905 F033
>>
>>
>

--
Zhang Yining
URL: http://www.zhangyining.net | http://www.yining.org
mailto: yining@zhangyining.net | zhang.yining@gmail.com
Fingerprint: 25C8 47AE 30D5 4C0D A4BB 8CF2 3C2D 585F A905 F033

Hello,

I have had some serious problems with YSlow since I started using it but
I wonder if it's only happening to me as I see no mention about it on
the list.

1st problem: I'm trying to get stats about this website:

	 http://gamersmafia.com/

But whenever I try to access the "Performance", "Stats" or "Components"
tabs in Firebug it hangs at the "Getting component information". If I
press on another YSlow tab I can get the results but there is missing
information: In the stats tab I get the total number of items of each
type (HTML/Text, IFrame, etc) but the size for all of them is 0.0K. If I
try to go to the "Components" tab I get results but no gzip/size
information.

2nd problem: Sometimes when I wait until the page is completely loaded
and then a few seconds more, I go check the Stats and Components tab and
if I refresh I get a different result. It's not just one more image or
one less image, it's a difference between 50 components and 140 components.

I am using latest Firefox 2.x official release.

Could anyone please check if you can get Firebug to work with this
domain and does any of you get inconsisten results regularly? I have
notified the YSlow team through the notify bug form some weeks ago but I
have yet to hear an answer from that direction.

Thank you,

--
Juan Alonso

I've experienced yslow hanging on a different site that had a flash
piece. With the piece removed, yslow behaved normally.

I've also seen your second problem. If I clear cache and pull up a page,
I'll get different stats than if I just refresh.

-----Original Message-----
From: exceptional-performance@yahoogroups.com
[mailto:exceptional-performance@yahoogroups.com] On Behalf Of Juan
Alonso
Sent: Thursday, April 10, 2008 4:00 AM
To: exceptional-performance@yahoogroups.com
Subject: [exceptional-performance] Problems with YSlow

Hello,

I have had some serious problems with YSlow since I started using it but

I wonder if it's only happening to me as I see no mention about it on
the list.

1st problem: I'm trying to get stats about this website:

	 http://gamersmafia.com/

But whenever I try to access the "Performance", "Stats" or "Components"
tabs in Firebug it hangs at the "Getting component information". If I
press on another YSlow tab I can get the results but there is missing
information: In the stats tab I get the total number of items of each
type (HTML/Text, IFrame, etc) but the size for all of them is 0.0K. If I

try to go to the "Components" tab I get results but no gzip/size
information.

2nd problem: Sometimes when I wait until the page is completely loaded
and then a few seconds more, I go check the Stats and Components tab and

if I refresh I get a different result. It's not just one more image or
one less image, it's a difference between 50 components and 140
components.

I am using latest Firefox 2.x official release.

Could anyone please check if you can get Firebug to work with this
domain and does any of you get inconsisten results regularly? I have
notified the YSlow team through the notify bug form some weeks ago but I

have yet to hear an answer from that direction.

Thank you,

--
Juan Alonso

------------------------------------

Yahoo! Groups Links

Thanks for the info Scott, at least now I can get the tabs working if I
disable the flash.

In case it helps the developers to fix the bug: the flash at
gamersmafia.com first makes an xml request and then retrieves a second .swf.

Scott Coldwell wrote:
> I've experienced yslow hanging on a different site that had a flash
> piece. With the piece removed, yslow behaved normally.
>
> I've also seen your second problem. If I clear cache and pull up a page,
> I'll get different stats than if I just refresh.

--
Juan Alonso

I've been trying to implement the new "Flush the buffer early [server]" best
practice.
However, with Apache and mod_gzip it seems that flushing the buffer early and
actually have
it flush doesn't work unless I set

mod_gzip_item_exclude file flush_test.asp

I had hopes of setting

mod_gzip_dechunk              No

making this work but to no avail.  Anyone out there been able to have their cake
and eat it to,
i.e., have both performance best practices work together?

Thanks for any help.

Hi guys,

Initially 13, then 14, now http://developer.yahoo.com/performance/ has
34 performance best practices (rules). These will gradually find their
way into YSlow, at least those that are testable.

I hope you'll find some interesting ideas to help you improve your page
loading speed today. Any comments and feedback appreciated. Let's make
the web a better place! ;)

Best,
Stoyan

Hi Stoyan,

Excellent news! I'm glad to read that. Is it possible to test or have
access to nightlies of YSlow?

Stoyan Stefanov wrote:
> Hi Juan,
>
>
>
> The first problem is fixed in the next YSlow minor release (0.9.5.beta -
> will be out really soon, probably next week)


--
Juan Alonso