C9 page load time problems - is it just me ?

seems like around the same time as mix the time it takes to load pages esp. the front page have went from semi-slow to amazingly horrible.

often i have to close and re-open ie before i even get the home page.

seems like the first time i try it just hangs ... i may get the page background and the silver community bar and the rest waits for stuff.

even starting this thread took 3 cylces of opening ie  to get here.

this happens on two different pc's that are on two different internet provider networks.

other web sites load normally.

this has been an on/and/off problem with C9 for over a year but the last 2 weeks seem much worse than i have seen in a long time.

i know that MIX added some traffic and that some chnages were made to file locations and the cdn useage.

something in all that has made the home page really suck

PS: location Tampa, FL -  ISP's are Verizon FiOS 5/15Mbps and BrightHouse / Roadrunner 2/10Mbps

I suspect the hamster is getting old and rundown

Charles said:

I'm seeing this randomly as well... Duncan has been made aware. Could be an issue with one of the nodes in our cluster.

C

It's been like this over 2 years now, just so you know

Charles said:

W3bbo said:

*snip*

Well, to be fair, it's been randomly like this for a long time. It's not always slow...

C

Why is C9 clustered? It's not like the site is serving more than a million pageviews a day, that's well within the capabilities of a modern single web server (assuming you have DB on a separate server). I'd only use clustering as a CDN for static content (like the videos and site artwork), but that shouldn't affect the performance of the page itself.

Charles said:

W3bbo said:

*snip*

There's this little thing known as reduncancy... Further, C9 traffic is quite high... We have the typical mutiple-server setup that most modern websites employ. The performance issues experienced here are a combination of inefficiencies that Duncan et al are ironing out. Right now, C9 is blazing for me. But not too many minutes ago, it wasn't. We are well aware of the problems and working hard to remove the bottlenecks (not all of these are code related.....).

C

It's politics, right?  Tell me it's politics. 

Charles said:

W3bbo said:

*snip*

There's this little thing known as reduncancy... Further, C9 traffic is quite high... We have the typical mutiple-server setup that most modern websites employ. The performance issues experienced here are a combination of inefficiencies that Duncan et al are ironing out. Right now, C9 is blazing for me. But not too many minutes ago, it wasn't. We are well aware of the problems and working hard to remove the bottlenecks (not all of these are code related.....).

C

C9 has NEVER been blazing for me, it varies from mediocre at its best, to totally unusable at its worst... and unfortunately, it is more often the later than the former. Personally, I'd suspect a routing & load-balancing issue... but it should not be hard to profile the problem.

Charles said:

W3bbo said:

*snip*

There's this little thing known as reduncancy... Further, C9 traffic is quite high... We have the typical mutiple-server setup that most modern websites employ. The performance issues experienced here are a combination of inefficiencies that Duncan et al are ironing out. Right now, C9 is blazing for me. But not too many minutes ago, it wasn't. We are well aware of the problems and working hard to remove the bottlenecks (not all of these are code related.....).

C

If it's just redundancy you're after, you don't need that kind of clustering for a solution. Just have an active/failover 2-node configuration. Whilst this would be a waste of computer resources initially, you could probably find something useful for the failover server to do, like CDN.

Having 5 minutes' downtime every month for updates is perfectly acceptable for us users. If you have a single DB server then the site remains live but you can easily show friendly messages.

This isn't rocket science. It's computer science

ScanIAm said:

Charles said:

*snip*

It's politics, right?  Tell me it's politics. 

I'm pretty sure it is. It MUST be something else than just "tech" involved here. I mean, it's not like C9 is amazon.com *LOL*, just a forum, some videos here and there and they haven't been able to "fix" the speed problem for 2+ years? Sheesh, it's almost sad.

How about everyone start a money bomb to donate some money so poor Microsoft can buy some better servers? hehhe... pun intended.

Charles said:

W3bbo said:

*snip*

It's more than this... We in fact do need more than 2 web servers. We have more than 2 web servers. We serve very large numbers of responses to requests originating all over the planet. It's not as simplistic as you make it seem, W3bbo....

 

C

Ok... yada yada yada, but so what?
Why should we care as end users? We are getting a crappy experience and that is what we see.

C9 isn't the only "big" forum with heavy traffic out there. There is absolutely nothing special ( technically speaking ) about C9 which would make such poor experience acceptable for the end users.

Fix the problem.

CKurt said:

W3bbo said:

*snip*

Try channel9 in IE9, its blazing fast

Oh, C9 is the first site ever to REQUIRE multi-cores CPU, and modern GPU? HAHAHA

Charles said:

turrican said:

*snip*

We are.
C

Managed to get couple load time spikes on IE9's network capture:

Id URL Method Result Type Sent Received Start (ms) Taken (ms) Wait Request Response Gap
11 http://channel9.msdn.com/ScriptResource.axd?d=vCwYTAQaeE_RVmvujYGGbqQ7URKM3u6yV-RnnIf3YFItsmxUHhwU9IotPfVxBMSSZNMxKIxQJL58YdK92xYeFMkqHE_rRcz90&t=2570d4ca GET 200 application/x-javascript 1,16 KB 97,37 KB 827 28 502 827 156 27066 453 27643
Id URL Method Result Type Sent Received Start (ms) Taken (ms) Wait Request Response Gap
19 http://channel9.msdn.com/Services/Tags.asmx/js?d=20100324020925 GET 200 application/x-javascript 1,07 KB 8,54 KB 30 264 54 772 30264 0 24508 0 1373

CKurt: So if 30 seconds to finish loading a page is Blazing Fast then sure it's Blazing Fast.