Slashdot Mirror

← Back to Stories (view on slashdot.org)

BBC on Website Slow Downs

Posted by ryuzaki0 on from the sure-gets-tricky dept.

HiveMaster writes "The BBC is carrying a report about the impact on websites as people try to get news regarding the war in Iraq. It talks of a report from Keynote Systems, which tests the reponsiveness of websites, which shows that the BBC news site has shown a fourfold increase in response times. However, Government sites in both the US and the UK are being hit, with the US Army site taking over 80 seconds to load at peak times." Also, here is a press release this. You can also read My journal where I've talked quite a bit about what Slashdot has done in preperation for traffic bursts.

13 of 206 comments (clear)

  1. Re:God bless the Internet! by at_kernel_99 · · Score: 4, Interesting

    Curiously enough, Pravda also has a story on the subject.

  2. Comment removed by account_deleted · · Score: 4, Interesting

    Comment removed based on user account deletion

  3. Ace's Hardware does it best. by BoomerSooner · · Score: 4, Interesting

    They run the whole thing off of a Sun Blade 500MHz with 2GB RAM.

    Pretty cool actually.

  4. Fourfold!?? by TrevorB · · Score: 4, Interesting

    Let's see. Loading the news.bbc.co.uk from Western Canada, right... now!

    10 seconds to render the whole page. OK, that's been significantly faster than I've been experiencing the last few days, It's been about 30-40 seconds in some instances in the past several days.

    Do try the low graphics version of the BBC, it loads almost instantly, and you can click on "Low Graphics" version while the rest of the page is trying to load.

    CNN does seem significantly better than years ago during major events. They must have tackling the planetary event slashdot effect thing down. But then again, I voted "Any non-us venue" on the poll.. :)

  5. How ironic... by meringuoid · · Score: 4, Interesting
    ... Slashdot making preparations for traffic spikes. I think we richly deserve to get massacred for days - shall we say ten minutes of downtime for every website we've blasted off the net over the years?

    ISTR that the NYT websute switched to a very graphics-light format in 11 September 2001, in order to cope with the mass traffic. Slashdot is already mostly text, but if necessary it could be lightened a bit. But I imagine the main load is CPU and memory, handling all those database queries and updates, rather than bandwidth; I don't see an easy way of dealing with that short of adding a few more machines. Imagine a... No, I won't :-)

    --
    Real Daleks don't climb stairs - they level the building.
  6. Light pages by rednaxel · · Score: 5, Interesting
    This shows the importance of light pages. The wide use of broad band has led many to ignore old guidelines. There's a lot of bloated pages with tons of ill-optimized pics and unnecessary gadgets.

    According this, the average web page is around 90 Kb. Google is a little over 10 Kb.

    --
    If you can read this, thank an english teacher.
  7. Iraq too by GiMP · · Score: 3, Interesting

    Iraq's website, www.uruklink.net, has been inaccessible quite often.. during the few occassions it has been online, it has been terribly slow.

  8. Re:Hmmm... by Kircle · · Score: 5, Interesting

    nope, but an interesting side effect could probably be more ad impressions and as a result, more revenue. honestly, /. could easily get more advertising impressions if they started temporarily mirroring the content from the stories posted on slashdot. everyone knows about the slashdot effect. why not use that in a manner where both slashdot and slashdot readers benefit?

    --

    -- Kircle

  9. Also a problem for places outbound connections by Sabalon · · Score: 3, Interesting

    My boss came to me around the 13th asking what were we gonna do on the 17th if we went to war...basically, how are mission critical apps gonna communicate over our school's internet connection if everyone is streaming video of war coverage?

    Luckily Shock and Awe started after most of our classes were done for the week and Thurs wasn't that bad. I guess with all the Kazaa traffic, streaming web didn't stand a chance :)

    Can't wait for that packetshaper to get here.

  10. Comment removed by account_deleted · · Score: 3, Interesting

    Comment removed based on user account deletion

  11. see bbc's own website stats by Anonymous Coward · · Score: 4, Interesting

    If you want to see oodles of stats about the BBC's own website, take a gander at http://support.bbc.co.uk/support

    Lots of mrtg graphs, response times, uptimes etc. Even a webcam of the support team :-)

  12. Re:Put a proxy-cache in front of it. by BrookHarty · · Score: 3, Interesting

    IF your data can be cached. Same thing with slashdot, the webpage is rendered depending on your karma, and account. Cache wont work, when its re-rendered for each person. (It should work for all the AC's loading the same page thou...)

    We use cheap netra t1's in a cluster behind load balancers, but our bandwidth usage per user is very small (1-5K) compared to Slashdot 50-100K page pulls.

    Strange how people design networks and server configurations without knowing the purpose. This is why when a product is handed off to the customer, they customers end up redesignning the architecture. Its not a cookie cutter world people...

    -
    You know the world's gone mad when the best rapper is a white guy, the best golfer is a black guy, the Swiss hold the America's Cup, France is accusing the USA of arrogance and the Germans don't want to go to war.

  13. DOS attacks on financial sites by Space+cowboy · · Score: 4, Interesting

    We host a fair few (60 or so) financial-orientated websites, with an average query-level of some 10-20 queries per second on the database supporting those sites.

    We have an 8-way cluster of machines to support this (way-overkill for the most part) but recently, we've been (almost) hitting limits... The apache service has logged peaks of 1000 connections/second, with the DB query-level going as high as 70,000/second....

    I'm actually fairly happy that the system can more-or-less cope with the load, but nonetheless, I want to make sure (or at least as-sure-as-possible) that we can't be easily DOS'd, so this weekend I'll be writing an Apache module to monitor the number-of-connections-per-second on an IP-by-IP basis, and take a decision to run a script depending on thresholds....

    I think stateful firewalls could probably manage it but for historical reasons we're stuck with what we have, and having apache call a bandwidth-limiting script on an IP address that's registered 5000 hits in the last minute (for example) seems reasonable :-)

    If there's something that can do this already, I'd like to know - I've found (ntal), but running a script per packet doesn't appeal :-( I prefer the idea of hitting a limit in Apache that triggers a script that limits access (dynamic firewalls)

    Ideas gratefully received :-)

    Simon

    --
    Physicists get Hadrons!