Title: Fast Running Blog database server uptime Post by: Sasha Pachev on January 13, 2010, 07:41:34 pm The MySQL database server on the back end of the Fast Running Blog has been up for over 306 days without a restart (means it has not crashed or got itself into an inconsistent state that would require a restart that entire time). Average load - 20 queries per second. Total queries - over 494 million. Database stats: total data size - 435 M, largest table - 379,129 records. 174,033 training entries.
OS load average rarely over 0.1, meaning the load could increase 10-fold before the users will notice load-related delays. Hardware: 1-CPU Pentium 4 2.4 GHz processor, 1 GB of RAM, 500 GB IDE drive. Everything except HDD was purchased for $65 from Geeks.Com. The HDD was about $90, I believe. The original RAM turned out to be bad, but fortunately I had some spare RAM laying around. Title: Re: Fast Running Blog database server uptime Post by: Steve P on January 13, 2010, 10:05:14 pm Congrats! That is much easier said than done.
You have a wealth of data that could be possibly be tapped for research purposes. Title: Re: Fast Running Blog database server uptime Post by: Paul Petersen on January 14, 2010, 07:08:15 am That's pretty cool. Just curious, what kind of bandwidth do you have?
Title: Re: Fast Running Blog database server uptime Post by: Jon Allen on January 14, 2010, 07:55:30 am Wow, impressive. Nice to know we can have 10 times more traffic without problems. And 20 queries per second- that seems like a lot to me.
Thanks for doing it, Sasha. Just curious, what happens if the power goes out at your house? Title: Re: Fast Running Blog database server uptime Post by: Tom on January 14, 2010, 08:21:37 am As a big computer geek myself I can appreciate those stats Sasha. Very impressive, nice job.
Title: Re: Fast Running Blog database server uptime Post by: Sasha Pachev on January 14, 2010, 03:45:55 pm Jon - I have it on a UPS that can last 10-15 minutes. The iProvo/Broadweave fiber-optic modem is also on its own UPS. So we can handle short power outages, and we have not had a long one for while. It is noteworthy that the server did not have to go down when they did the electrical work this summer when we added a dining room (the Lord poured out so many blessings that there was not room enough to receive them).
I wish I could say that all those 20 queries per second were human generated. If they were, the blog would finally get a nice UI as I would have the budget for it :-) Most of it is bot traffic. I suspect due to a high level of page linking and the abundance of pages the bots really like the site. Some buggy ones get stuck in a loop. But it gives a nice stress test for the site. Paul - the connection is 20 MBps from iProvo/Broadweave. $43 a month. The bandwidth is definitely underutilized spiking only to maybe 100 KBps occasionally. As much as people complain about iProvo and the government trying to compete against private business, I think it was a good thing. In this case the private business (Qwest and Comcast) got lazy. Now that they have been given a kick in the behind, they are offering high bandwidth at a reasonable price as well instead of complaining that it is not feasible. And to add a bit of a flame to the fire, I do not consider corporations of that size a true private business. Once an organization gets big enough it tends to lose the competitive edge of a private business that make the free market economy thrive. Title: Re: Fast Running Blog database server uptime Post by: Paul Petersen on January 14, 2010, 03:57:43 pm That's amazing bandwidth for the price. Qwest charges about $500/month for a dedicated T1. Yeah, it duplexed and 99.9% reliable, but the price and the speed is pathetic.
BTW - thought you might find this interesting: http://www.letsrun.com/2010/09letsrun0113.php RunningAhead is #16, and they are a dedicated online running log. When will the Fast Running Blog crack the Top 20? Title: Re: Fast Running Blog database server uptime Post by: Steve P on January 14, 2010, 09:24:23 pm I pay $40 for 1 Mbps. :'( And because my HOA signed an exclusive agreement, it's my only option (until WiMax becomes a reality). $43 for 20 Mbps sounds excellent.
Title: Re: Fast Running Blog database server uptime Post by: Sasha Pachev on January 15, 2010, 11:42:54 am Paul - we would need to increase our traffic by a factor of 8.
Title: Re: Fast Running Blog database server uptime Post by: Eric Day on January 15, 2010, 12:38:00 pm How many times do I need to login to help with this? ;D
Awesome up-time ! And $43 for 20MBps bandwidth? wow, I pay something around $75 for 2 mbps... (third world country with cost of living of better than first world country). Title: Re: Fast Running Blog database server uptime Post by: Maurine Lee on January 18, 2010, 11:52:40 am Those are some awesome statistics. I have to reboot my SQL Server several times a week to deal with a memory leakage we haven't been able to trace yet.
Title: Re: Fast Running Blog database server uptime Post by: Sasha Pachev on January 18, 2010, 12:05:09 pm Maurine:
A while ago if you did a Google search for MSSQL it would come back with "Did you mean MySQL?". MySQL team had a lot of fun with it. Google actually uses MySQL for the backend of AdWords. Title: Re: Fast Running Blog database server uptime Post by: Jon Allen on January 19, 2010, 07:47:56 pm Sasha- you might have jinxed yourself. When I open the FRB in Chrome or Firefox, I get the following errors:
Warning: session_start() [function.session-start]: Cannot send session cookie - headers already sent by (output started at /home/asksasha/web/running-log/blog.php:314) in /home/asksasha/web/running-log/blog_entries.php on line 408 Warning: session_start() [function.session-start]: Cannot send session cache limiter - headers already sent (output started at /home/asksasha/web/running-log/blog.php:314) in /home/asksasha/web/running-log/blog_entries.php on line 408 Warning: Cannot modify header information - headers already sent by (output started at /home/asksasha/web/running-log/blog.php:314) in /home/asksasha/web/running-log/blog_entries.php on line 409 Warning: Cannot modify header information - headers already sent by (output started at /home/asksasha/web/running-log/blog.php:314) in /home/asksasha/web/running-log/blog_entries.php on line 410 And when I try to login, I get the following: Warning: session_start() [function.session-start]: Cannot send session cookie - headers already sent by (output started at /home/asksasha/web/running-log/blog.php:314) in /home/asksasha/web/running-log/trainlog2.php on line 804 Warning: session_start() [function.session-start]: Cannot send session cache limiter - headers already sent (output started at /home/asksasha/web/running-log/blog.php:314) in /home/asksasha/web/running-log/trainlog2.php on line 804 Warning: Cannot modify header information - headers already sent by (output started at /home/asksasha/web/running-log/blog.php:314) in /home/asksasha/web/running-log/trainlog2.php on line 472 Title: Re: Fast Running Blog database server uptime Post by: Sean Sundwall on January 19, 2010, 09:37:26 pm I'm getting a similar problem. cant post my run today
Title: Re: Fast Running Blog database server uptime Post by: Sasha Pachev on January 20, 2010, 08:55:47 am Fixed. Apologies for the problems. Note that the database server was still up, otherwise you would not have been able to post to the forum. I just made some changes in the PHP code and forgot to test them fully before making them live. Nice to know that the blog is being used, though.
Title: Re: Fast Running Blog database server uptime Post by: Kory Wheatley on January 20, 2010, 01:04:21 pm I've had a Dell Dimension 4100 Pentium 3 for 10 years and have not had one problem with it. I've upgraded from Windows ME, to Windows 98 Special Edition, and then to XP. The only thing I've changed is I installed a upgraded processor, bigger hard drive. The power supply is the original one and the system board is as well.
|