I started questioning the objectivity of Netcraft stats lately and wanted to share my opinions and get some feedback about this...
I need a reliable web hosting plan (not dedicated, co-hosted, etc.) and trying to find a hosting company which I can depend on. For this, I have been doing some reliability tests (pinging + checking pop3 + http download) on several hosting companies from different locations within/outside USA (West, East, Nortwest, London, etc.). I did these every once in a while using my+my friends' personal DSL connections, and also tried some commercial services such as atwatch.com in April. According to all of my tests (mine + the commercial sites I used), he.net is definitely the most reliable one.
However, when you look at various rankings on Netcraft web site (i.e. uptime.netcraft.com/perf/reports/performance/Hosters?tn=april_2004&reverse=0), he.net is always worse then the ones I tested myself, for example in the April report, he.net is #29. According to netcraft failure rates, it is worse then the following sites that I tested myself: jumpline, ipowerweb, hostway.
Once again, I tested these from various sources, and he.net is far better than these 3 in terms of reliability. In fact,there are cases where he.net never failed where others were failing right and left. And, looking at the postings on this site, everyone acknowledges he.net's reputation as well.
Please don't get me wrong here, I am not doing any advertisement for he.net. In fact, I find their pricing model ridiculous (the $9.99 plan offers only 30Meg space, and the next one is $24 and offers 1GB) and trying to avoid getting a service from them. THus, I have been trying to find something that is comparable to them. But, unfortunately, I could not manage to find anything yet that can compete with he.net in reliability. And, these forums are killing me, cause everyone is promoting some other site. For example, lunarpages.com has very positive feedbacks here, but according to results I got, it is way way worse then many others in terms of reliability...
So, my question is simple, what are these guys at Netcraft smoking But, seriously, is netcraft another one of those sites where these companies pay to get high spots on rankings? How reliable are these results?
Also, is there any other site that does this kind of comparison?
You said yourself, you did tests every once in awhile using DSL lines while they monitor uptime and traffic for multi-million dollar companies. No offense meant, but I would tend to think they are the more reliable guage. In all honesty, your tests might be a good indicator of the number of problems your ISP has in routing you to those sites/companies that tested poorly for you.
Check out thehostingchart.com, they do a chart like netcraft I believe.
I am not sure if you read all of what I wrote (i.e.: "I tested these from various sources")...
If it was only my DSL line, I might have agreed with you. But as I said, I have been testing from totally independent backbones with different DSL/Cable providers. On top of that, I tested from relatively reliable university networks, etc. Additionally, I also used commercial services like atwatch.
If one of them, or a few of them were indicating the contradiction I mentioned, I would still agree with you. But when all of them are pointing in the same direction (which is what I explained in my previous message), unfortunately your comments don't make much sense.
To leave aside all these, come on... Can a company like jumpline compete with he.net in relibility? Or even worse, IPowerWeb which provides hosting for $9 is ranked as more reliable in these stats...
As I mentioned in my previous e-mail, I just want to get a feeling of what is going on, and wanted to share my thoughts with people, so they would not be mislead as I had been.
I think I misunderstood that you tested several hosting companies from your friends and yours personal DSL accounts, plus a few commercial services (of which you mentioned one). I see now that you meant you tested from all different locations, not were testing hosting co's in different locations. Sorry.
I don't see where you said you used several different ISPs to test. But I do have a question: did you testing and the commercial services you tried show the same results?
Don't get me wrong, I'm not knocking you. I just thought the post was a little vague on the details of how you tested in comparison to how netcraft tests, which is what I would like to know before using them for anything. If they are inaccurate people should be aware!
I did not want to get into details and bore people, but to be more precise, I performed two tests: One from a DSL line (West Coast) and one from another DSL line in NYC. Additionally, I performed tests from atwatch.com, webmetrics.com, 1stwarning.com and elkmonitor.com.
As for the results, as you can imagine, they are not exactly same... But as I said, they pretty much all point in the he.net direction. Just to give you an idea:
1stWarning.com: lunarpages uptime rate was around 99.8%, 1 jumpline site: 100%, the other: 90%, He.net: 100%, ev1servers.net: 99.5%.
And, for the HE.net and the 100% jumpline sites, He.net was around 2 times faster than jumpline.
1 lunarpages site is 99.7%, while the other is 100%. both Jumpline sites, HE.net and ev1servers were also 100%, and HE.net was the fastest among them.
HE.net was at least 3 to 9 times faster than others............. These are the only results I could pull up from my files, and could not find the results for my own tests, but they were similar to these.
Shortly, in all of them HE.net was considerably better than others according to my results.
So, it is pretty clear that quality and price are directly proportional, but as I said, all I am trying to do is to find a hosting option that is comparable to HE.net in terms of reliability. That is all...