Saturday, November 21, 2009

Web Host

I've been running a variety of low-data, low-traffic websites since 1999. I've been with three web-hosts, one of which sucked. My requirements are:

  • good server up-time
  • good server response-time
  • php
  • mysql
  • add-on domains

I'm currently with (as of Nov 16th 2009), and I'm very happy. They have a great price and a great service. And they're Canadian, which is nice. I paid $159.46 CAD (all-in) for one year with unlimited bandwidth, 600GB storage, unlimited add-on domains, unlimited mysql dbs, cpanel, and a bunch of other frills. They seem like a good company with a good history. And the best part is that their server is lightening fast. After switching to iweb, there was a significant and obvious improvement in page load times. Here's my pingdom data so far.

pingdom uptime chart for iweb pingdom response time chart for iweb

100% uptime and 248ms average response time is plenty good for me. My previous host, seemed pretty good at first. I paid $182.16 CAD (all-in) for one year of hosting with plenty of bandwidth, plenty of storage, plenty of add-on domains, 3 mysql dbs, cpanel, and a bunch of other frills. My reason for moving to midPhase in the first place was the add-on domains (more on that below), but after I'd signed up and seen cpanel for the first time, I was hooked. Easy sign-up, great admin interface, midPhase was looking considerably superior to my previous host. But the response-time sucked. At first I thought it was my fault, bad scripts or something, maybe my internet connection. So I signed up for a pingdom account (great service by the way) to monitor both servers. And it turns out, yes midPhase response time sucked, but it wasn't just that. The really slow loads ~1min or so were actually mini-downtimes. Have a look at the charts.

pingdom uptime chart for midphase pingdom response time chart for midphase

So, the 608ms average response time is pretty sad, but it's the sheer number of down-times that I couldn't tolerate. Granted they're short. (I think actually shorter than 5min, that's probably just the pindom resolution). But I'd be working away using the punchclock, and bang!, the site would be dead for two minutes. That's a real pisser. And don't think I was measuring some resource intensive page. These stats are from a simple html file on the primary domain. In fact, before getting the pingdom account, I put up simple php, html and pdf files on the primary domain to rule out specific problems with dns, db, php, and file serving when one of the mini down-times was being observed.

After gathering some data, I went to midPhase support about it. Here's how it went.

From: Michael Holtstrom, Mar/11/2009 8:15:54PM

I've noticed that from time to time, my site appears to be down for approximately 1 minute. (i.e. I get the ...not responding... page)

Feb 23, 2009 ~7:30am
Feb 2, 2009 12:42pm
Jan 21, 2009 10:41am

Since I'm not using the site all the time, I must assume that actual downtime is much frequent.

Is this expected?
Do you provide any uptime guarantee?
Can you suggest any actions to diagnose the problem? (i.e. logs, etc)


From: James Rasmussen, Mar/11/2009 8:22:48PM


I am showing that your website is loading, resolving, and up at the moment. You can try looking at the error logs in cPanel to see if there is anything there of import to your uptime.

From: Michael Holtstrom, Apr/13/2009 7:59:47PM


As I said earlier, I've noticed that from time to time, my site appears to be down for approximately 1 minute. (i.e. I get the ...not responding... page)

I've been using pingdom to monitor it for the last month. Screenshot is attached.

Is this expected? It seems unusual to me that the server goes down so often. For example it was down four times on April 11th. This hasn't been my experience with other hosts.

Do you provide any uptime guarantee?


From: James Rasmussen, Apr/13/2009 10:12:43PM


It looks like that was from october, november, and december. Could you post anything more recent? Your uptime is 99.52%, which is quite good.

From: Michael Holtstrom, Apr/14/2009 6:33:35AM

Please take a closer look. That log is from April 9th, 10th, 11th, 12th of 2009.

The date format is day-month-year.

Yes, 99.52% sounds quite good, however my past experience with servers has been characterized by long (days, weeks, etc) periods of up-time.

Your server is non-responsive for several short periods every day.

That sounds like a problem to me.

1. Is it expected that your servers become non-responsive several times every day? What is the reason for this?

2. Do you offer any uptime guarantee?


From: Tami Mitchell, Apr/14/2009 1:47:01PM

Hello Michael,

I'm sorry that you are frustrated with the service. The hosting account you are currently on is on a shared server, and due to the nature of our shared servers there may occasionally be downtime due to excessive load. Please note that we do not offer an up time guarantee. If you feel that you require more control over your server, you may want to take a look at our dedicated hosting plans. This way your service will only be based on your site, not the activity of multiple clients.

From: Michael Holtstrom, Apr/16/2009 9:34:21AM

Thank you for your reply.
I don't mean to cause any trouble.
In many respects I prefer your service over other hosts I've used in the past.

In regards to "there may occasionally be downtime due to excessive load". I assume that the load is relative to the number of accounts you host on a single machine. For April 1-15, I observe an average of 1.8 downtime events per day. I thought this was unusually high and that perhaps there might be an issue with this machine which should be addressed.

Is it your policy to load your machines to this level?
If so, I can accept that.

In regards to "we do not offer an up time guarantee".
Can you explain what is meant by the attached "99.9% uptime guarantee" from your linux-hosting page.

The site in question is low-traffic/non-profit, so I don't think that dedicated hosting is the right fit for me.

From: Tami Mitchell, Apr/16/2009 3:36:47PM

Hello Michael,

I would like to apologize, I had not recently checked if we offered an up time guarantee as I was unaware that we had one. This was certainly an error on my part, and I'm sorry for the wrong information.

While we do try to utilize our shared servers as much as possible, what you describe (1.8 downtime events/day) should not be happening and is not satisfactory performance. We are not aware of any issues on esc124, but I will let our admins know to keep an eye on it in the future as there may be some problem accounts on it.

Thank you for letting us know.

Well that's the story. So when my year's hosting ran out, I looked around and settled on iWeb, who as I've said are great. Before all this, I was with for nine years, and I still use them for all my domain registrations. They're also a Canadian company and provide very reliable service. Have a look at the pingdom stats below. In the nine years, over two separate hosting accounts, I only ever noticed one downtime. They had some kind of catastrophic failure and the server was gone for most of the day. They sent out an apology email and said that they'd had to restore it from the previous day's backup. I did have some other minor problems over the years (like a bad server clock), but their support staff resolved them all.

The only reason I left was for add-on domains. They only let you have one domain per hosting account. If you want other domains to point to a sub-folder on your server, you have to fake it with a service like or So I switched hosts. And I'm glad I did. Having cpanel and filemanager is awesome. And I think that iWeb will continue to be a high quality, affordable web host.

pingdom uptime chart for iweb pingdom response time chart for iweb
{ "loggedin": false, "owner": false, "avatar": "", "render": "nothing", "trackingID": "UA-36983794-1", "description": "Advice on web hosts. Bad service from midphase. Better service from iweb.", "page": { "blogIds": [ 13 ] }, "domain": "", "base": "\/michael", "url": "https:\/\/\/michael\/", "frameworkFiles": "https:\/\/\/michael\/_framework\/_files.4\/", "commonFiles": "https:\/\/\/michael\/_common\/_files.3\/", "mediaFiles": "https:\/\/\/michael\/media\/_files.3\/", "tmdbUrl": "http:\/\/\/", "tmdbPoster": "http:\/\/\/t\/p\/w342" }