Actually this is a cache machine, so it is not hosting a site it is caching for 3 towns. It is going to be a cluster in a couple months, this started as a small project and then we found it was saving us around 25% of our bandwith and speeding up our service up to 40% so we have kept moving more to this server and I think I am around the max it is going to do. I want to get it moved to a 3 server cluster.
The server actually is not the big for hardware, it is a P4 2.4 GHz with 1 gig ram and 4 20 gig ide drives, one per controller with an ide controller card. It is running redhat 9.0 with all apps compiled for the machine to get a bit more proformance. Also I spent a good bit of time getting the drives to work faster, I have them up to 47 Megs/S Transfer which is good because I have up to 1000 files being opened and written to each drive at a time so around 4000 files being saved and opened at any time. |
cool read of a puerto rican internet provider that does something the same that triples the bandwidth of connections (i am sure they can be looked up, that was over 2 years ago), how does the performance throughput go after all that up time?
on a side note, i know all about hdparm, what settings u give it and how much u increase their performance? |
Actually the funny thing is that the longer the server stays up the better the performance gets :-) The best way I can explain that is it keeps fine tuning what it puts in memory and where to find files on the drive.
As for hard drive performance, first make sure you have the 7200 rpm drives and a really good 133 ulta ata cable, only put one drive per controller, and check with the manufacturer on the specs of the drive, most times you can find the right values to give hdparm for a drive right from the manufacturer. Right now on the machine I am running around 55-60% hit rate on my cache, it is a 30 gig cache, and anything comming off the cache is super fast, I can do a 40 meg windows update in about 35 seconds. Which is nice as I used to spend most of my time downloading the updates but they are all cached now. |
Quote:
Thanks man. Appreciate it :) Peace, Whitehat |
As of right now:
Quote:
In fact, it's a laptop; IBM Thinkpad X30. Once a time, this was a really expensive machine. Anyways, as you guys can see, it treats me well. |
On my old rig, I had this at one time.
Code:
root@smoker / # uptime :D :D |
My box used to be regularly up for 1 month or so. Now I decided to save on electricity so I switch it off every night.
|
The current longest uptime on a server I use is:
Code:
$ uptime |
One of the guys in my LUG found a server in his company's server room (it had been sitting there quietly serving without any trouble) that had been up for over three years.
My personal best is something like 156 days for my webserver when I used to self-host (Slackware 10.x on a IBM PC 300). |
My desktop is sort of a electricity hog (read: lots of fans, blue LEDS, overclocking, noise, sex appeal) so it gets turned off whenever I'm not using it.
My EEEpc usually gets suspended whenever it's not in use, and I think my record was about 26 days uptime with Slackware 13.1 before the battery ran out on it.. :o |
i never gave it a single thought !
I often go for months at a stretch.... is there any world record for the same ? |
Quote:
I have seen posts from people tho of several years. I went 242 days once. Hurricane Katrina sort of put my lights out. That was with a Abit mobo and a AMD CPU. :D :D |
Just a waste of energy.
Kind regards |
Quote:
Some of the systems with such long uptimes are servers. They are supposed to be up all the time. That's why they are servers. :D :D |
I have mentioned this on another thread, but I built a SuSe file server for my kids school. The network maintenance got taken over by some company so I handed over the passwords and forgot about it. After what must be 4 years I got asked by one of the teachers to have a look at it because the company of so called experts, after charging for the servers maintenance for years, suddenly decided "they didn't have the expertise". (And that rip off is a sore point for another thread)
It was just the HD was full so I tidied up a bit and added another one to the LV. So at least 4 years, at a conservative estimate, the server was filing and printing with no intervention whatsoever. Pretty cool. |
All times are GMT -5. The time now is 12:17 PM. |