On 4/18/12 2:45 PM, Roberto Galoppini wrote:
Hi all,
we have been testing AOO downloads for about 10 days, below our findings:
- downloads per day ( http://goo.gl/jKbnC )
- info about geographies (
http://sourceforge.net/projects/openofficeorg.mirror/files/stats/map )
- operating systems (
http://sourceforge.net/projects/openofficeorg.mirror/files/stats/os )
We spent the first days to tune our delivery system and facilities to
serve AOO downloads, We actually started by putting a CDN in place
that could grab files from our master mirror as its origin server, and
we learned we needed to pre-cache files to speed up downloads. Once
CDN was working properly, and we got those test results we updated the
test to servey we eventually switched to serve up these files from our
normal mirror network. As of today we updated also the download page,
now reporting the information formerly delivered by the download page
at Apache (e.g. http://goo.gl/WXSuk).
Our understanding based on the above data, and the expectations set by
the community prior to the test is that we are not serving all
downloads, (e.g. downloads.openoffice.org/es, automatic updater
downloads, etc) and it's unclear how many downloads are actually
served by different pages. We should probably try to route all
download traffic, so that we can provide more accurate download stats.
We have tested our expanded mirror network, and burst CDN capacity,
and are confident that we are well positioned to serve much heavier
download traffic than is expected at release time. Since the traffic
will be high also be high for updates, we might take additional
precautions and continue to route auto updater downloads through
MirrorBrain, and use Apache and sourceforge.net mirrors as a backup.
To the extent that sourceforge.net is the primary recomended download
source for AOO downloads we feel it is important to keep providing the
PPMC with accurate stats info (see above), also via API, as mentioned
before (see http://goo.gl/tTgZ6). If there is more download
statistical data that the PPMC would like to see, please let us know
so that we can try to meet those needs.
Hi Roberto,
thanks for this useful numbers and the good report. Very impressive
numbers. I am looking forward to see some numbers for AOO 3.4 ;-)
Juergen
Roberto
On Mon, Apr 9, 2012 at 9:08 PM, Roberto Galoppini<[email protected]> wrote:
On Fri, Apr 6, 2012 at 6:59 PM, Rob Weir<[email protected]> wrote:
On Fri, Apr 6, 2012 at 10:34 AM, Rob Weir<[email protected]> wrote:
Please note: https://issues.apache.org/ooo/show_bug.cgi?id=119194
I'm about to apply this patch. It will cause download requests to go
to the SourceForge mirrors instead of MirrorBrain. We want to confirm
that things are set up properly on the SF side, before we get the
deluge of downloads for AOO 3.4.
So please respond to this post with any oddities you see or hear about
with users downloading OOo 3.3 releases over the next few days.
This is live now (in production).
Two issues that I see so far:
1) The MirrorBrain downloads were redirecting the user to
http://www.openoffice.org/download/contribute.html, along with URL
parameters that indicated the file that was downloaded.
This is very useful for a couple of reasons:
A) This is our best chance to make the user aware of the support
options, the mailing lists, the install instructions, etc. It is our
opportunity to "hook" them into getting more involved with the
community and the ecosystem.
B) The URL parameters give us information on what platforms and
languages are downloaded most often. For example:
http://www.openoffice.org/download/contribute.html?download-mirrorbrain&files/stable/3.3.0/OOo_3.3.0_Win_86_install-wJRE_en-US.exe
The current SourceForge download path does not do these things. I
think this is important and needs to be addressed. Maybe something
could be done by forming similar URL's and showing that page in an
IFrame from the SF download page?
We can put links to AOO contribution resources next to the ad unit. We are also
open to help to find new developers through our media channels, and we have an
'help wanted' feature that can be used for this scope.
2) I have not done any timings, but the download speed seems slower,
at least the time taken until the download is actually started.
Think this was on Friday when we were still tuning our mirrors and caching
files.
Roberto
-Rob
-Rob
====
This e- mail message is intended only for the named recipient(s) above. It may
contain confidential and privileged information. If you are not the intended
recipient you are hereby notified that any dissemination, distribution or
copying of this e-mail and any attachment(s) is strictly prohibited. If you
have received this e-mail in error, please immediately notify the sender by
replying to this e-mail and delete the message and any attachment(s) from your
system. Thank you.
