[AusNOG] Anyone else notice the click frenzy traffic

[Anthony Spruce]

Looking further into it, there was no Obvious Caching on the front-end, nor
were they utilising gzip encoding, which was pushing up page-size.



I’d be interested as to what was congested considering over PIPE there
doesn’t seem to be an abundance of traffic, however there was excessive
load times on that link on any site including the hosts.



By now, they would’ve likely lost a large portion of the original hits in
any case, so the Amazon thing will end up being too little too late.



Kindest Regards,
Anthony Spruce

Technical Support
*
Servers Australia Pty Ltd
*
Fax:                        (02) 4307 4201
Network Ops:       (02) 9037 4343
Web:                      *http://www.serversaustralia.com.au
*Office Address:    2/2 Teamster Close, Tuggerah NSW 2259
Postal Address:    PO Box 3187, Tuggerah NSW 2259





*From:* ausnog-bounces at lists.ausnog.net [mailto:
ausnog-bounces at lists.ausnog.net] *On Behalf Of *Sean K. Finn
*Sent:* Tuesday, 20 November 2012 9:47 PM
*To:* Joshua D'Alton
*Cc:* ausnog at lists.ausnog.net
*Subject:* Re: [AusNOG] Anyone else notice the click frenzy traffic



Did anyone consider CDN for pics? Srsly,even 250,000 people hitting at
10kbps is 2,500,000,000 bps or 2.5Gbps. 10kbps might only barely suffice
for static HTML at that rate.



An apache web server starts to struggle at 500 SIMULTANEOUS requests,
meaning 250,000 visitors would need to be load balanced across 500 server
instances, assuming your LBs can handle that amount of sessions, keep track
of the backend servers, which likely multiplies the traffic to double the
request amount, and that is just to ensure a 10kbps experience without any
pictures, and static HTML/js.



500 servers running any type of PHP or ASP or god forbid anything Java and
forget it.



CLICK CLICK BOOM



More than likely Telstra and Optus saturated someone's interconnects,
there's not too many people with lazy gig links hanging spare about the
place.



S.


On 20/11/2012, at 8:02 PM, "Joshua D'Alton" <joshua at railgun.com.au> wrote:

Indeed, even something as simple as cloudflare can turn a fail-site into a
world-class one with a few clicks. Obviously not the case for every site,
but even just the caching provided that reduces network utilisation (which
I believe is what is the cause of this outage).

On Tue, Nov 20, 2012 at 8:47 PM, Martin Barry <marty at supine.com> wrote:

$quoted_author = "Matthew VK3EVL" ;

>
> I wouldn't say it's dirt cheap. High volume websites do cost a bit
> to run properly.

I think the point was that they probably haven't even started with the
basics, all of which are cheap or free if you have already built out the
rest of the infrastructure. Along the lines of: preferring things to be
static or cache-able; switching off heavy features under load...

cheers
Marty

_______________________________________________
AusNOG mailing list
AusNOG at lists.ausnog.net
http://lists.ausnog.net/mailman/listinfo/ausnog



_______________________________________________
AusNOG mailing list
AusNOG at lists.ausnog.net
http://lists.ausnog.net/mailman/listinfo/ausnog
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ausnog.net/pipermail/ausnog/attachments/20121120/e929f19e/attachment.html>