
06-30-2009, 06:42 PM
|
|
Newbie
|
|
Join Date: Jun 2009
Posts: 8
|
|
What are the beneifts of RAM ?
I've just upgraded my server on Liquid Web:
Processor: Pentium 4 3GHZ Hyperthreaded
Memory: 4GB DDR SDRAM
HDD 160G Sata
Centos 5
My question is, what's the benefits of RAM to my websites in comparison to CPU.
Also as a client of Liquidweb for over 6 months, I've experienced a lot of downtime, hence why've I upgraded to 4GB Ram, but I just want to know how this help.
Thanks 
|

06-30-2009, 06:49 PM
|
|
iNET Interactive
|
|
Join Date: May 2006
Location: San Francisco
Posts: 7,154
|
|
Everything on the server uses RAM to run and when you run out of RAM, it starts swapping (using the hard drive as memory) which isn't good.
However, did you ask LiquidWeb why you were experiencing lots of downtime?
|

06-30-2009, 06:51 PM
|
|
Junior Guru
|
|
Join Date: May 2009
Location: Morrison, Colorado
Posts: 190
|
|
RAM allows you to processes more information at a time before writing it to a disk. If someone is processing information from a PHP script, the information is stored in RAM while being processed by the CPU. Just to name a few uses. Do you know exactly what is causing the downtime? Is it their connection or your server for sure? There are a number of causes to a server crash and not all of it can be solved by adding more RAM. What are you using the server for?
|

06-30-2009, 06:57 PM
|
|
Newbie
|
|
Join Date: Jun 2009
Posts: 8
|
|
I own a network of 4 large forums (mainly vbulletin), they get around 4k uniques a day each.
The server always crashes due to high CPU load, in which I have to send a ticket for them to restart apache, but that's all they do, there's no long term solution.
So I spoke to a senior system admin and he suggested upgrading from 1GB to 4GB but I just hope it prevents the downtime, esp since Liquid web claim to offer 100% uptime...
This is screenshot of my dashboard in PIMs (Liquid web account panel):
ooj.org/view/dowtime29.jpg
The red lines represent server crash, the thicker the line the longer the crash was for...
|

06-30-2009, 09:24 PM
|
|
Newbie
|
|
Join Date: Jun 2009
Posts: 23
|
|
Potter, sorry to jump in here, but i am starting a new business and about to use a shared pro account from liquidweb to start with, i obviously won't have many visitors to start with but hopefully will be able to grow, can you tell me more about liquidweb? what is your opinion about them now? thanks in advance and sorry for changing the subject (i just want 1 post response from someone who has dealt with them first hand  ) thanks in advance
|

06-30-2009, 09:35 PM
|
|
Newbie
|
|
Join Date: Jun 2009
Posts: 8
|
|
If your sites are not established and don't experience much traffic then any web host is good really.
Liquid Web are great for support and value for money.
Because my sites are big and active, I get a lot of downtime as I said in my first post.
Though I've just upgraded to 4GB on my dedicated server so this should solve.
In terms of growth, depending on how fast you grow you might as well host with them now as they are pretty great though I hear great comments from Rackspace
|

06-30-2009, 09:42 PM
|
|
Web Hosting Master
|
|
Join Date: Dec 2003
Location: Montreal, Canada
Posts: 697
|
|
Let us know the results of your uptime after the upgrade! Have you considered upgrading to a more high performance server altogether, or are you simply seeking for a quick-fix that will reduce downtime.
|

07-01-2009, 10:38 AM
|
|
Junior Guru
|
|
Join Date: May 2009
Location: Morrison, Colorado
Posts: 190
|
|
You may want to consider upgrading the processor if that's the case. You may have enough RAM but not enough computing power as I know vBulletin is very resource intensive. Look into some Xeon plans if available and I you're problems will most likely go away. Hosts guarantee uptime in accessibility to servers (ie connection). If you are crashing the server, then the guarantee usually doesn't apply (unless you are on a managed machine or have a clause in your contract).
|

07-01-2009, 11:45 AM
|
|
New Member
|
|
Join Date: Jun 2009
Posts: 4
|
|
I would think you get more mileage out of tuning your setup. High cpu load shouldn't kill processes. Overcommitted memory because apache is starting too many children and triggering the OOM killer would get you downtime. If you are running with the default setup, MaxClients is probably set way too high for the amount of memory you got. Anyway with a single SATA drive, I think you will most likely be I/O bound before cpu. But really gather some metrics when your server is under load and see where the bottleneck is. Its better than randomly upgrading things and hope that helps.
|

07-01-2009, 01:30 PM
|
|
Retired Moderator
|
|
Join Date: Oct 2002
Location: EU - east side
Posts: 21,920
|
|
Quote:
|
Though I've just upgraded to 4GB on my dedicated server so this should solve.
|
If your server was lacking in RAM, it would lead to swap space being used intensively which would slow the server down, and could eventually lead to things breaking down entirely.
I don't suppose that the tech recommended such a significant RAM increase on a whim. It would make no sense.
Quote:
|
esp since Liquid web claim to offer 100% uptime...
|
That is the network uptime SLA ( https://www.liquidweb.com/about/dedicatedsla.html). A host cannot guarantee 100% server/website uptime, especially if it turns out that the machine you bought is not enough for the job at hand. They're not miracle makers.
Let's hope that the memory increase is all that's needed to bring you stability. As the others have said though, it is possible that while it was the most significant bottleneck at this time, other hardware or software improvements may be necessary.
|

07-02-2009, 03:00 PM
|
|
Newbie
|
|
Join Date: Jun 2009
Posts: 8
|
|
I just upgraded to this plan now as the CPU was still high:
Processor: Dual Xeon 2.8GHZ Hyperthreaded
Memory: 4GB DDR Registered/ECC
Hd1: 160GB 7200RPM SATA / 8MB Cache
Hd2: 160GB 7200RPM SATA / 8MB Cache
RemoteBackup: No Remote Backup Needed
OS: Linux - CentOS 5
(+64 Bit)
ControlPanel: CPanel / Web Host Manager
(+ServerSecure)
(+Fantastico/XController)
Bandwidth: 4000GB Monthly Transfer (2000 in + 2000 out)
|

07-02-2009, 03:29 PM
|
|
Junior Guru
|
|
Join Date: May 2009
Location: Morrison, Colorado
Posts: 190
|
|
Let us know if that solves the issue, quite curious now. Looks like an awesome machine so I doubt you will have problems with it for what you need it for.
|

07-02-2009, 03:44 PM
|
|
WHT Content Curator
|
|
Join Date: Nov 2007
Location: Dallas, TX
Posts: 8,410
|
|
Quote:
Originally Posted by redrocksdatacenter
Let us know if that solves the issue, quite curious now. Looks like an awesome machine so I doubt you will have problems with it for what you need it for.
|
I am also curious. Please do keep us updated. That is quite a CPU boost (nice!).
-mike
__________________
Mike G. - Limestone Networks - Account Specialist
Dedicated Server Hosting - Premium Network - Passionate Support - Fusion Reseller Platform
Dallas Datacenter - Unmanaged and Managed Services - Intel Ivy Bridge Now Available!
@LimestoneInc - Dedicated Server - 877.586.0555 x1
|

07-02-2009, 06:03 PM
|
|
iNET Interactive
|
|
Join Date: May 2006
Location: San Francisco
Posts: 7,154
|
|
Quote:
Originally Posted by redrocksdatacenter
Let us know if that solves the issue, quite curious now. Looks like an awesome machine so I doubt you will have problems with it for what you need it for.
|
There's a chance that the CPU upgrade might not have been enough. I'm assuming those are Netburst Xeons which were basically P4s.
|

07-02-2009, 07:42 PM
|
|
Powered by Inve :D
|
|
Join Date: Aug 2007
Posts: 1,809
|
|
How many physical cores do you actually have? Is it just a dual core xeon (2 cores) or it dual processors (single core each)?
How are you looking at the CPU usage, are you running a top and looking at the load? Correct me if I'm wrong, but IOwait time (hard drive load) can affect the numerical number reported on top.
With active forums, I would think you'd have a lot of data going in and out. There could be a chance that your harddrives aren't fast enough. I noticed you had two, are these RAID1 or RAID0? If you can afford it, I might go for a Yorkfield Xeon, which is 4 cores (quadcore), with 4GB of ram, and two 15K SAS drives in a RAID1.
I'm not sure exactly where the problems lie, but if it's server related (as in not powerful enough) the above server would probably fix it. I believe it would be faster in all areas, CPU, probably memory, and definitely on the HDD side.
__________________
:: Bobby B - InveHosting.com - High End Hosting Solutions
:: Shared, Reseller, VPS, Dedicated, and more
:: Check out our many options and addons at InveHosting.com!
:: Become a fan at Facebook.com/InveHosting for exclusive specials!
|
Related posts from TheWhir.com
|
| Title |
Type |
Date Posted |
| Thread Tools |
Search this Thread |
|
|
|
| Display Modes |
Linear Mode
|
| Postbit Selector |
|
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
|
|
|
| Login: |
|
|
| Advertisement: |
|
|
| Web Hosting News: |
|
|
|