Hello,

My current hosts ping is 10 pings; average ping time = 0ms

The host I am considering to move to, is 10 pings; average ping time = 65ms (and I checked a couple sites that they host and those sites are: 10 pings; average ping time = 16ms)

I'm using this utility to check the ping: http://webservices.cnet.com/ping/

My Question is : Is the difference between 0ms and 16 or 65ms that big of a deal?

And at what point would you say a ping really sucks? (or what is a reasonable ping time?)

Also are there any other tools online that you can check the integrity of a server? And what do you look for in particular?

Thanks for the reply!

Ping should not be a HUGE thing. I mean once hosting company could have much more accounts, and the server could just have more load. Also I do not know how cnet works, if they ping from multiple hosts or not. So lets say cnet was located in New York. Host A was also in new york getting a 10 ms ping. Lets say Host B is in Cali, it will have a slower ping due to the distance, dosent mean its slower.

Use tracerroute instead that way you know what route is being used and where the server actually is. It is a better way of comparing because it is closer to the apples to apples scenario.

If you do a search on the board for this test you will see that there are some concerns about this particular perfmorance measurement test (mostly because of the lack of disclosure on part of CNET as to how the measurement is done, and because the other hosts compare yours to may be advertisers).

I also don't think that 0ms measurement is accurate. It is simply impossible to get a 0ms response from a server in another data center, even on the same backbone.

To give you an example, if I ping the router at the edge of my data center (2 hops away, just on the other side of the firewall), I get the following response:


10 packets transmitted, 10 packets received, 0% packet loss
round-trip min/avg/max/mdev = 0.833/7.848/67.285/19.815 ms


Now, you could argue my host sucks and their network is crap. On my home network, I have 5 machines connected to a Cisco 2924, which has a 3 Gig backplane, this is my response when I ping from one machine to another:


10 packets transmitted, 10 packets received, 0% packet loss
round-trip min/avg/max/mdev = 0.476/0.562/0.815/0.096 ms


So, my response is still almost 1ms for a machine on the same network.

Supposedly CNET was able to gain an AVERAGE of 0ms to your host from 10 different backbones?? I don't buy it :D.

To answer your question, I would say you should look for hosts who have less than a 200ms response time, but I would also do it for a longer period of time. One snapshot does not give you enogh data.

If you really want to see how well a potential host performds, use one of the free monitoring services for a week or so, and see what type of response you get:

http://www.qwkmon.com
http://www.netmonitor365.com/

Originally posted by DigitalXWeb
Use tracerroute instead that way you know what route is being used and where the server actually is. It is a better way of comparing because it is closer to the apples to apples scenario.

I disagree with this. The traceroute will show you the path takes to a host from your location. Obviously you are more interested in how well the rest of the world can get to your server (unless you are the primary visitor to your site). You may have bad connectivity to the Internet, so the long route may not be indicitive of general performance. A traceroute will also not determine response time, only the route.

OK well what do you guys do to check the integrity / speed of a server you are interested in?

I used this: http://visualroute.visualware.com/

And the host I am considering gave these results:
13 hops, Starting in Dulles, VA with Verio, then Dallas to UUnet, then washington DC UUnet then ended up in Baltimore, MD
TTL = 245

My current host - (actually my own website) did this:
17 hops, starting with Dulles, VA with Verio, then Washington, New York, Chicago with Cogent Technologies then ended with my host in South Lake Tahoe. TTL = 246

So how does that look? Are the 13 hops that much better than the 17, or are they both pretty much the same quality?

Thanks for the reply!

That CNET service doesn't tell much about quality of hosting, you should not rely on it in choosing a web host. And also, visualroute is no use either. What you should do is measuring the download times from a lot of different ISP, depending on your target audiences, if you have any. Ping times or hop count won't tell you which one is better.

I wish CNET could take down its ping service :(.