Moved to Linode.com

For many years, aplawrence.com ran on a BSD based VPS (Virtual Private Sever - sharing by virualization) server. I was quite happy with that, but a few years back the hardware it was running on started becoming unstable and as the hosting company I was with no longer offered VPS servers, they wouldn't move it to new hardware.

I tried Rackspace. They were expensive and after crashing within a few days of trying it, I gave up on them. I also tried Amazon S3 - cheaper than Rackspace, but they really seem to go out of their way to make everything confusing. I also didn't like that they charge for extreme usage rather than just limiting you - a DOS attack could cost a fair amount of money before bringing me down!

I am unwilling to pay for dedicated hosting, so I reluctanly settled for shared Linux hosting at the same host where the BSD VPS had been. It was reliable, but I really disliked not having control. No root access, no ability to run jobs with "nohup" (they'd just get killed after an hour). I couldn't run "screen", couldn't install software, couldn't serve compressed html .. none of that is expected with shared hosting, of course, but I missed it.

Linode

I heard about Linode rather accidentally and immediately liked what I saw. They are very reasonably priced - I went with the 768MB package for $29.95 a month with no contract (you can prepay to get cheaper prices). They have a wide choice of Linux distros to choose from, a comprehensive help system and a simple, yet very powerful command interface. It's all Xen powered, very easy to add storage space, additonal RAM and so on. You can choose from multiple hosting locations, too.

I'm paying just a few dollars more than I was for shared hosting, but have full control again. It's also much faster:


# old system
$ time find . -name "*.html" | wc -l
11181

real	0m0.625s
user	0m0.040s
sys	0m0.039s

# Linode system
time find . -name "*.html" | wc -l
11181

real	0m0.040s
user	0m0.019s
sys	0m0.020s
 

This was AFTER changing DNS, so the Linode was serving pages while that ran; the old system had already dribbled down to a few hits from China now and then.

I have far more disk space than I need:

Filesystem           1K-blocks      Used Available Use% Mounted on
/dev/xvda             12106720   1411344  10197304  13% /
devtmpfs                370876       116    370760   1% /dev
none                    371084         0    371084   0% /dev/shm
none                    371084        48    371036   1% /var/run
none                    371084         0    371084   0% /var/lock
none                    371084         0    371084   0% /lib/init/rw
/dev/xvdc             12106720    722360  10886288   7% /dup
 

The system has been generally loafing:

Linode usage graphs

I am very happy to be back on a VPS!



Got something to add? Send me email.





(OLDER) <- More Stuff -> (NEWER)    (NEWEST)   

Printer Friendly Version

-> -> Moving my site to a XEN VPS at Linode.com


5 comments



Increase ad revenue 50-250% with Ezoic


More Articles by

Find me on Google+

© Anthony Lawrence







Thu Jun 16 12:31:20 2011: 9562   TonyLawrence

gravatar


Comments should be working again now. I had done some clean up and rearranging of data and accidentally killed comments. Should be OK now.





Thu Jun 16 12:34:52 2011: 9563   TonyLawrence

gravatar


Please do notify me of any broken or misbehaving items, thanks.







Mon Jun 20 15:21:16 2011: 9578   Ralph

gravatar


Thanks Tony for your aricle about linode.com as a hosting provider.
As I am running my VPSs in Europe I became curious to see how good
a VPS performs in comparison to the big players.
Obviously VPS are not playing in the same field, but are they
much slower for hosting a website? I tackled this question with a
performance test and threw 10 requests at a number of sites using

httperf --server bigname.com --num-conns=5 --num-calls=1

The results were a big surprise (for me):

Here are the main characteristics:

RANK    SITE           connections/s      Time/conn    Network I/O 
1   cern.ch            7.7    7.7       129.8 ms        3.5 
2   securesolutions.ie 4.3    4.4       230.8 ms        39.0 
3   bing.com           3.5    3.6       285.2 ms        1.7 
4   facebook.com       3.3    3.6       299.8 ms        1.6 
5   google.com         3.3    3.3       300.5 ms        1.9 
6   kerry-linux.ie     3.3    3.2       300.7 ms        59.2 
7   nasa.gov           3.3    2.8       303.2 ms        58.7 
8   apple.com          3.2    2.4       313.1 ms        0.9 
9   hp.com             2.7    2.8       372.5 ms        1.4      
10  youtube.com        2.2    2.7       447.9 ms        1.2 
11  aplawrence.com     2.2    2.2       451.5 ms        1.3 
12  schneier.com       2.0    2.0       501.9 ms        1.1 
13  microsoft.com      1.1    1.3       894.4 ms        0.6 
14  twitter.com        0.9    0.8      1140.9 ms        45.0 

Well, Tony your new node is not far away from Youtube and HP.
And even Nasa and Google is only slightly ahead. But what knocked
my socks off was that two of my sites came in second and sixth
(at least after a reboot) and only the elementary particle guys
that invented the web were clearly ahead of the crowd.

Not too bad, eh?

Ralph






Mon Jun 20 16:37:03 2011: 9579   TonyLawrence

gravatar


No too bad at all. Though you'd need to test pages with the same mix of graphics/css et al. and the same number of GETS required to really judge this. For example, you could (temporarily) steal a random page from hete and put it at the same directory depth at one of your sites and then compare.

But yeah, Linode is pretty good :)



Mon Jun 20 18:18:34 2011: 9580   Ralph

gravatar


Taking a closer look at the original httperf data, I found that indeed the various sites send a different amount of bytes back as a reply. This is the size of the replies in bytes:

1) cern.ch                  409 
2) securesolutions.ie      9152      
3) bing.com                 441 
4) facebook.com             415 
5) google.com               511 
6) kerry-linux.ie         18177 
7) nasa.gov               18177 
8) apple.com                216 
9) hp.com                   478 
10) youtube.com             484 
11) aplawrence.com          556 
12) schneier.com            506 
13) microsoft.com           540 
14) twitter.com           52317 

As some sites return a substancial bigger reply than others, their ranking would probably increase if all sited returned the same size of packets. This is true particularly for twitter, nasa , securesolutions and kerry-linux.

Kind regards

Ralph






------------------------
Kerio Samepage


Have you tried Searching this site?

Unix/Linux/Mac OS X support by phone, email or on-site: Support Rates

This is a Unix/Linux resource website. It contains technical articles about Unix, Linux and general computing related subjects, opinion, news, help files, how-to's, tutorials and more.

Contact us





FORTRAN's tragic fate has been its wide acceptance, mentally chaining thousands and thousands of programmers to our past mistakes. (Edsger W. Dijkstra)

A man can be destroyed but not defeated. (Ernest Hemingway)







This post tagged: