homepage Welcome to WebmasterWorld Guest from 107.20.109.52
register, free tools, login, search, pro membership, help, library, announcements, recent posts, open posts,
Pubcon Platinum Sponsor 2014
Home / Forums Index / Google / Google SEO News and Discussion
Forum Library, Charter, Moderators: Robert Charlton & aakk9999 & brotherhood of lan & goodroi

Google SEO News and Discussion Forum

    
Dedicated vs. Shared server: Effect on google traffic
Marvin Hlavac




msg:4058671
 10:24 pm on Jan 11, 2010 (gmt 0)

I'm thinking about switching from a shared to a dedicated server.

Google Webmaster Tools reveals that 75% of websites on the web are faster than mine, and that even after I reduced size of images, and removed unnecessary items. I even got rid of Google Analytics code, just to save every unnecessary KB.

There may not be much more I can do, and I'm starting to seriously consider abandoning the inexpensive shared server plan, and moving to a fully managed dedicated server at the same hosting company. It will cost me around $200/mo.

Questions:

1. If the dedicated server is going to be at the same hosting company, can I realistically expect to gain better speed? Or should I search for a hosting company that has faster internet connection? How would I know which host is faster (and overall better)?

2. I would like to hear from people who made the move from shared to dedicated. Can you share your story? Did it help, or not really? Did Google give you more traffic just because of the move from shared to dedicated?

My site started just over two years ago, and the traffic has been increasing in a surprisingly steady manner. Last month (Dec 2009) the site had around 83k unique monthly visitors. This level of traffic should happily live on a shared server, but I'm just hoping to improve the speed. Will a dedicated server achieve the desired speed increase?

 

devil_dog




msg:4059030
 12:50 pm on Jan 12, 2010 (gmt 0)

Generally speaking, dedicated servers would be faster than shared server cause you have all the resources to yourself.

But in your case, when you say "Google Webmaster Tools reveals that 75% of websites", it seems you are talking about client side performance which does not have a lot to do with your server speed.

My suggestion would be to install Firebug plugin in Firefox and observe the Net tab (screenshot at [getfirebug.com...] ). If you notice long waits for your server to send response, you know you have a server side issue where a dedicated server might help.

Google WMT tells that my site is slower than 94% of the sites... thats probably cause i have so many 3rd party widgets and ad codes on my pages. My site is functional, i.e. the main content is visible within 1 second(note i am on a slow network with high latency) of loading... yeag all the ads n sharing widgets take a little longer but thats acceptable to me, it should be acceptable to Google as well, Google aint so dumb, I wouldn't read too much into these stats, just nice to keep it as low as possible without hindering the functionality.

Marvin Hlavac




msg:4059144
 3:21 pm on Jan 12, 2010 (gmt 0)

devil_dog, thanks very much for pointing me the right direction. I've installed the Firefox plug-in. However, I'm not certain I'm able to tell if a server is fast or slow just by using this tool to observe the "Waiting for Response" times. The times vary quite a lot each time I refresh a page.

The "Waiting for Response" times I get can be anywhere from 20ms to 4 seconds, no matter which site I access.

I tested my site, I tested also webmasterworld.com, and other sites, but I'm unable to tell which server is faster. How should I view the data, or how can one tell which site is on a faster server?

mslina2002




msg:4059161
 4:05 pm on Jan 12, 2010 (gmt 0)

I am actually in the same situation as you and have also been researching VPS, dedicated and cloud hosting.

My site is less than 2 years old and starting to increase in traffic to about 75k visitors. So similar stats.

I noticed that the site was very slow towards the end of last year and found out on Google WMT that my site was about 95% slower than most sites. Eek. It's a WP site so a started deleting plugins I did not absolutely needed.

Installed FF Firebug plugin and started tweaking images, javascript, gzip wherever I could, cacheing, etc. The site is now a little faster and down to 85%. Still not good.

My host suggested that I have a professional look at the scripts and if that didn't help to move to private or semi-private hosting. They do not offer dedicated servers so it meant that I would have to leave them. I love their excellent service so if I leave, I hope to find the same kind of service elsewhere.

It has become an obsession to try to get this number down because I know visitors are clicking away and that conversions are down. Since I have tweaked the site and my site has become it a bit faster, I definitely see an increase in conversions.

I have read on here from some forum members have benefited from dedicated servers by seeing increased conversions and I would bet that it helps with SERPs as well if a site is faster.

smallcompany




msg:4059168
 4:18 pm on Jan 12, 2010 (gmt 0)

Be careful with jumping onto what WMT says.

Between sites that I have in WMT (around 30), those that are on shared are in a green area (fast), while those on separate VPSs are around 50%-70% slower than the rest.

Those on VPS are sites with heavy traffic and therefore more chance to get various client sides measurements as many people are visiting every day.

Just switching to the new server (whatever hosting option you pick) may not do anything.

On the other side, there are cases where a webmaster is supposed to switch anyway, simply because the traffic has matured, no matter what WMT says.

P.S.
Also see what's the accuracy. Google clearly says if the data is of low, medium, etc accuracy and gives some data points. It's in bold.

encyclo




msg:4059219
 5:48 pm on Jan 12, 2010 (gmt 0)

To find the answer, you need to do some content analysis to find out what exactly is making your site slow. Analytics code is minuscule and not a real issue, removing it should not make much difference.

What is your content? Plain HTML, videos, lots of graphics? Are you making calls to external resources (third-party code, addons, etc.)? How heavy are your pages (combined HTML, CSS, graphics and JS)?

devil_dog




msg:4059264
 7:07 pm on Jan 12, 2010 (gmt 0)

If you are worried that your server is slow, look at the speeds in :-
WMT > Diagnostics > Crawl stats

See the last graph titled "Time spent downloading a page (in milliseconds)". Now if thats slow, then moving to a bigger server might help. If thats within acceptable limits (my avg is 244ms im happy with anything < 400ms) then you apparently dont need to switch.

The "Labs > site performance" has more to do with the client side programing... Sure your server speed is one of the factors involved, but generally the important factors r the javascripts, etc...

Marvin: When you use the firebug tool, monitor only the requests to your server and in each individual request check purple portion of the bar. If thats long... even for static files... time to move on ...

Marvin Hlavac




msg:4059297
 8:09 pm on Jan 12, 2010 (gmt 0)

@ smallcompany:

see what's the accuracy. Google clearly says if the data is of see what's the accuracy. Google clearly says if the data is of low, medium, etc accuracy and gives some data points. It's in bold. and gives some data points. It's in bold.

I don't see the "low, medium, etc accuracy", but I do see the suggestions of what could potentially be improved. I will look into it if there is something I could do. Thanks!

@ encyclo:

What is your content? Plain HTML, videos, lots of graphics?

It's a forum, running vBulletin script version 3.8.4 with a few add-ins and modifications. A few pages do have pictures, but most do not.

@ devil_dog:

WMT > Diagnostics > Crawl stats

See the last graph titled "Time spent downloading a page (in milliseconds)". Now if thats slow, then moving to a bigger server might help. If thats within acceptable limits (my avg is 244ms im happy with anything < 400ms) then you apparently dont need to switch.

457 is my average. Would you consider switching?

mslina2002




msg:4059342
 9:28 pm on Jan 12, 2010 (gmt 0)

457 ms sounds pretty good to me. Have you tried the suggestion using the FF firebug plugin. They have quite a few ideas as to how to speed up your site.

Mine averages 1742 ms. So time to switch?

devil_dog




msg:4059705
 8:37 am on Jan 13, 2010 (gmt 0)

457 is my average. Would you consider switching?

457 is quite decent...
personally i somehow am not comfortable with shared hosting... its difficult to find out exactly where a bottleneck is (asuming there is one).

encyclo




msg:4059812
 1:12 pm on Jan 13, 2010 (gmt 0)

83k unique monthly visitors. This level of traffic should happily live on a shared server

a forum, running vBulletin script version 3.8.4 with a few add-ins and modifications

I would be uncomfortable having this setup on a standard shared server, as such equipment has the tendency to be oversold in terms of capacity - it all depends on what other sites are on the same server. However, as you say, you don't appear to be in dedicated server territory just yet. A decent VPS plan with guaranteed resources would probably be sufficient, you're looking at $25 to $50 a month in that case.

One issue not mentioned is local hosting - is your current server close to your users? ie. do you have a US-based server for European users or a similar scenario? If so, you might want to look for more local hosting which would reduce the latency.

Patrick Taylor




msg:4059824
 1:44 pm on Jan 13, 2010 (gmt 0)

On one of my shared server websites I noticed that dynamic pages with content served from a MySQL database were periodically quite slow, whereas flat HTML files on the same shared server remained fast. The problem was mostly cured by asking my host to move the database to another newer MySQL server with less load on it from other customers.

The first thing I would try is to ask your web host if they're aware of the issue.

Seb7




msg:4059953
 4:10 pm on Jan 13, 2010 (gmt 0)

Marvin, Make sure your website is fully optimised first before thinking of switching to a dedicated server. Check your HTML, style sheets, Java, image sizes, any other page objects.

Marvin Hlavac




msg:4060294
 12:05 am on Jan 14, 2010 (gmt 0)

@ encyclo:

A decent VPS plan with guaranteed resources would probably be sufficient, you're looking at $25 to $50 a month in that case.

I'm now exploring this option. My host charges $59/mo for the least expensive VBS plan. If I want them to manage it, I pay additional $45/mo for a "Fully Managed" service.

do you have a US-based server for European users or a similar scenario?

This is not an issue. The server is in the U.S., and most of my site visitors are in the U.S.

@ Patric Taylor:

The first thing I would try is to ask your web host if they're aware of the issue.

They seem to think everything is just fine on the server side. They actually don't think I will gain anything by moving to a dedicated or VPS.

@ seb7:

Marvin, Make sure your website is fully optimised first before thinking of switching to a dedicated server. Check your HTML, style sheets, Java, image sizes, any other page objects.

I've been checking everything to the best of my (very) limited abilities. I'm still reading on optimization of vBulletin forum software.

Thanks to all! Everyone has been most helpful.

Leosghost




msg:4060309
 12:31 am on Jan 14, 2010 (gmt 0)

My host charges $59/mo for the least expensive VBS plan. If I want them to manage it, I pay additional $45/mo for a "Fully Managed" service.
'KNL !

They seem to think everything is just fine on the server side.

I'll bet they do.. if their standard rates are reflected in their VPS rates .especially managed VPS at $45.oo per month on top of "basic" VPS basic plan rates of $59.oo per month!

They actually don't think I will gain anything by moving to a dedicated or VPS.

They probably dont want to scare away the golden goose ..or at least they dont want him comparing prices , specs and packages elsewhere ..

"managed" on a VPS should come free ..it is "virtualised" so the "managed" part is "money for old rope" ..

Look into what has been suggested already ..and look around ..

CainIV




msg:4060319
 12:50 am on Jan 14, 2010 (gmt 0)

Try using Code Speed in Firebug. Really helped me eliminate some pretty big but obvious problems.

I have never used shared hosting, optimized for a dedicated VPS system which is managed.

Marvin Hlavac




msg:4060347
 1:34 am on Jan 14, 2010 (gmt 0)

@ Leosghost:

Look into what has been suggested already ..and look around ..

I would love to look over suggestions of reputable hosting companies trusted by the members of Webmaster World. If there is a thread around here where hosting companies were discussed, please point me there...

Robert Charlton




msg:4060387
 2:52 am on Jan 14, 2010 (gmt 0)

from the Webmaster Business Issues forum Charter...
Public discussion and recommendation of hosting companies is no longer allowed at WebmasterWorld. We are closing the floodgates to get a better handle on the spam that always crops up in hosting discussions. The current threads will remain available to be read but there will be no new discussion permitted.

This is also covered in this forum's Charter [webmasterworld.com].

There is a Webhosting Issues and Options forum [webmasterworld.com] in the Supporters area where discussion of specific host companies is allowed.

Marvin Hlavac




msg:4064581
 10:33 pm on Jan 20, 2010 (gmt 0)

Robert, I understand the motivations behind such forum policy. This and other of your forum rules make this site a valuable source of info w/out spam.

Here's an update on my situation:

I've done my research over the past couple of weeks, and three days ago I switched from shared to VPS. I switched to a different hosting company, and the first two days my forum experiences technical difficulties which caused it to be down intermittently. That has been corrected. The load speed of pages that used to be 4 to 5 seconds is now 2 to 3 seconds.

If I observe an effect of this on Google traffic, I will post back with details at a later time.

Thanks everyone for sharing your valuable insights.

Robert Charlton




msg:4064662
 1:34 am on Jan 21, 2010 (gmt 0)

Marvin - Thanks for the update.

The load speed of pages that used to be 4 to 5 seconds is now 2 to 3 seconds.

I'm glad your load speed has improved. While I've seen many forum pages that feel like they're taking longer than 2 to 3 seconds to load, 2 to 3 seconds still seems slower than it should be.

Are the pictures you mentioned optimized? Sometimes images which are uploaded are straight from someone's camera, and that can put a considerable crimp in site speed.

Looking forward to further details re Google traffic.

Marvin Hlavac




msg:4064680
 2:10 am on Jan 21, 2010 (gmt 0)

Robert, most pages have no pictures actually at all. The site runs on a popular forum software, and while I have removed many unnecessary features from the pages, even the fastest forum pages still do not load faster than 2 to 3 seconds. I'm not sure if it is even possible to get better speed with the forum script used.

[edited by: Robert_Charlton at 7:55 am (utc) on Jan. 21, 2010]
[edit reason] removed specifics [/edit]

Robert Charlton




msg:4065510
 7:51 am on Jan 22, 2010 (gmt 0)

...I switched from shared to VPS.

While a VPS offers some isolation from other hosting customers in software terms, a VPS by definition is not a dedicated server. (Note that word "virtual" in there). You still share the hardware... and, in particular, the shared CPU and shared RAM could well affect site speed. This very much depends on the load the other customers hosted on your box place on the machine.

If you are seeing load times of 2 to 3 seconds, you should check at different times of day, and also have people you know check your site and give you feedback. And ask your host what kind of load the other sites on the box are placing on the system.

Marvin Hlavac




msg:4065765
 3:56 pm on Jan 22, 2010 (gmt 0)

Robert, thanks for the reply.

1. It's good to know there still is a way to improve the performance by moving from VPS to dedicated. It will likely be the next step, and it will likely happen within the next few months.

2. I wonder if perhaps I'm not using the proper way to judge the time. I've used two to three different sites that measure load time, and I've used also one web browser add-in, and they all report to me the approximate 2 second load time (despite some people telling me the page loads almost instantly). When I use the browser add-in to measure the load speed of even (WebmasterWorld) this page, after I do Ctrl+F5 to refresh this page, the tool also gives me the load time of almost 2 seconds.

dstiles




msg:4066037
 11:50 pm on Jan 22, 2010 (gmt 0)

Try adding a Timer call at the start and end of page parsing (assuming you are using dynamic page generation) and print the total parsing time on the page. This should give an idea of a) server cpu time used; b) efficiency of parsing your code; c) time to load "library" code inclusions. Check at peak and nadir activity times for your site's accesses.

Something to note: some browser plug-ins seriously slow down display time - eg HTML Validator, if it's turned on for all (or your specific) sites. Anti-virus software may also slug it - apropos which some firewall software pulls un-zipped pages so that could slow down delivery. Obviously rendering time is also affected by the computer that it's delivered to (CPU speed, CPU loading, what else is running in the background).

Broadband speed is often very optimistic, especially the high-speed ones: see if you can find the true speed. BT in the UK has just announced a super-speed fibre cable option but fails to mention that local delivery is still by copper and that some servers are just plain over-loaded in the first place - expect much slower speed than the ISP claims. They also fail to mention contention ratio (how many others are using the same "cable") and how many of them are also trying to download movies, BBC programs, videos, whatever - that can be a real slowdown.

Some pages from my own sites load and render within a second, even those with databases behind them. If I have (eg) an AV scan going on or a lot of Firefox extensions turned on then they are slower.

I accept your original request concerned google timing, but google isn't the customer and I personally wouldn't trust their time results: I know they are incorrect for me.

Marvin Hlavac




msg:4066463
 8:15 pm on Jan 23, 2010 (gmt 0)

dstiles, thanks for pointing out the browser add-ins slow things down. Indeed! Very good point! :)

SEOPTI




msg:4066688
 6:03 am on Jan 24, 2010 (gmt 0)

shared server = amateurs
dedicates server = pros

dstiles




msg:4066953
 11:05 pm on Jan 24, 2010 (gmt 0)

That is a very naive view.

shared server = businesses with limited resources, especially in the current financial climate.

dedicated server = businesses that make a very good income from their sites (or have enough money from other sources that they don't have to worry).

The vast majority of web site owners are amateurs in the field of web sites. They rely on professionals to tell them the best options FOR THEM.

All of my customers are on a shared server. I manage the dedicated server they share. I don't believe my customers suffer in any way from this.

dataguy




msg:4067099
 7:25 am on Jan 25, 2010 (gmt 0)

shared server = amateurs
dedicates server = pros

I'm with SEOPTI on this. I can't imagine allowing my web business to suffer at the whims of the 2 to 2,000 other web sites on the same hardware as my web site(s).

I know a guy who switched web hosts to get a faster connection to his server and it instantly increased his traffic and ad revenue to the tune of $200 per day. Even though the new hosting cost $300 per month more than what he was used to, I (I mean he) would never go back.

Have you played around with VMWare or Virtual Server? Sure it compartmentalizes your server software so that it can easily be backed up and moved to different hardware, but one VPS on a dedicated server can't be more than 20% as fast as a true dedicated server (that's 80% slower.) And we know that nearly all VPS hosting has more than one VPS on a physical server. Running VPS has got to be close to the equivalent of running a site on a 486 computer. Who would do that?

Global Options:
 top home search open messages active posts  
 

Home / Forums Index / Google / Google SEO News and Discussion
rss feed

All trademarks and copyrights held by respective owners. Member comments are owned by the poster.
Home ¦ Free Tools ¦ Terms of Service ¦ Privacy Policy ¦ Report Problem ¦ About ¦ Library ¦ Newsletter
WebmasterWorld is a Developer Shed Community owned by Jim Boykin.
© Webmaster World 1996-2014 all rights reserved