View Poll Results: Which is faster network (from your location)?

**alexs** · 10-31-2002, 02:56 AM

Why the traceroute is so long to reach jaguarpc?
Aletia old servers were faster from europe: 250ms vs. 630ms or more...

I hope jaguarpc will use another provider to better support europe connections.

Try to do some ping to hamster and to mercury at jaguarpc:
64.247.37.223 HAMSTER (aletia)
66.227.70.0 MERCURY (jaguarpc)

Check the time in "ms" that is higher and reply to the poll!

==========
ALETIA HAMSTER TRACEROUTE FROM ITALY/EUROPE

1,XX (masked)
2,XXX (masked)
3,XXXX (masked)
4,XXXXX (masked)
5,80.19.134.144,88ms,r-fi66-5-vl19.opb.interbusiness.it,----
6,151.99.98.85,88ms,r-fi63-fi66.opb.interbusiness.it,----
7,151.99.101.189,79ms,r-rm213-fi63.opb.interbusiness.it,----
8,151.99.29.139,50ms,r-rm83-vl3.opb.interbusiness.it,----
9,195.22.196.181,199ms,pal6-ibs-18adsl-it.seabone.net,----
10,195.22.209.233,284ms,linx-lon1-racc1.lon.seabone.net,----
11,195.66.224.94,284ms,core1.ltn.nac.net ,----
12,209.123.11.29,390ms,at-0-0-1-1145.gbr1.oct.nac.net,----
13,64.21.102.2,189ms,gi-1-1-95.msfc1.oct.nac.net,----
14,207.99.8.198,220ms,fa0-0.core1.pwebtech.com,----
15,64.247.37.223,250ms,ns1.aletia.com,----
==========

JAGUARPC MERCURY TRACEROUTE FROM ITALY/EUROPE

1,XX (masked)
2,XXX (masked)
3,XXXX (masked)
4,XXXXX (masked)
5,80.19.134.24,88ms,r-fi67-vl14.opb.interbusiness.it,----
6,151.99.98.81,91ms,r-fi62-fi67.opb.interbusiness.it,----
7,151.99.101.193,89ms,r-rm215-fi62.opb.interbusiness.it,----
8,151.99.29.150,90ms,r-rm180-vl3.opb.interbusiness.it,----
9,195.22.196.217,239ms,pa5-ibs-2adsllight-it.seabone.net,----
10,195.22.208.42,240ms,mil8-mil7-racc1.mil.seabone.net,----
11,217.147.129.133,241ms,sl-gw10-mil-12-1.sprintlink.net,----
12,217.147.128.38,240ms,sl-bb20-mil-8-0.sprintlink.net,----
13,213.206.129.25,240ms,sl-bb21-par-12-0.sprintlink.net,----
14,213.206.129.69,109ms,sl-bb20-lon-13-0.sprintlink.net,----
15,213.206.128.38,268ms,sl-bb21-lon-15-0.sprintlink.net,----
16,213.206.131.22,240ms,None,----
17,212.187.131.146,251ms,ae0-16.mp2.london1.level3.net,----
18,212.187.128.46,99ms,so-3-0-0.mp1.london2.level3.net,----
19,212.187.128.138,180ms,so-1-0-0.mp1.washington1.level3.net,----
20,209.247.11.169,380ms,so-0-2-0.mp1.houston1.level3.net,----
21,209.247.11.182,391ms,gige9-0.hsipaccess1.houston1.level3.net,----
22,209.247.121.218,251ms,unknown.level3. net,----
23,66.7.143.178,381ms,o0-2jp1.hou002bd01.yipes.com,----
24,66.7.143.165,390ms,o0-2bd1.hou002bd02.yipes.com,----
25,66.7.143.158,391ms,o0-2bd2.hou003bd02.yipes.com,----
26,66.7.143.54,390ms,o5-3bd2.hou005ap01.yipes.com,----
27,66.54.158.66,420ms,summittrading.com,----
28,66.227.76.108,630ms,spinelli-group.com,----
==========

Bye, Alex

**megmond** · 10-31-2002, 04:26 AM

Actually, Aletia's servers were apparently spread out over the US, so you might see various speeds and access times depending on the host you are referring to.

**alexs** · 10-31-2002, 04:35 AM

I supplied the IP of the old hamster, the traceroute is the same and the IP goes to the same place where the old hamster was, at Aletia.

IP addresses could not be reassigned so quickly. Hamster was moved just 2 days ago.

Alex

**jason** · 10-31-2002, 07:16 AM

When they move a site, the don't move the server immeadiately. The old "hamster" is pobably still where it was originally if sites were just moved from it. In a few days it will be taken down and shipped to Houston and will become a new server with a different name.

Every trace I've seen since the move has been going through Yipes, although I'm sure that there are other providers that they are using for redundancy. Maybe they can adjust their routing tables so that a different provider defaults to the European connections.

--Jason

**alexs** · 10-31-2002, 07:51 AM

To: Jason
========
I agree with you Jason.

I hope they'll optimize routing tables for better performance and minimum hops.

To: Forum Users
============
Anyway continue to vote and test both connections from your location!

You need only to go to command line and do 2 commands:
ping 64.247.37.223
ping 66.227.70.0

Bye, Alex

**lookout** · 10-31-2002, 08:28 AM

Well, from where I'm at, the new IP of 66.227.70.0 is almost twice as speedy as 64.247.37.223.

Makes sense that it would be like that for me in California. Hamster was in New Jersey. The new server is in Houston, which is another big hub, but considerably closer.

I have noticed that the new server itself (had been on Dragon, now Gold) seems a bit slower at processing requests. May still need some tweaking, although I wouldn't be surprised if it reflects the change from a pair of ultra scsi 160 hard drives to EIDEs. I don't have any empirical evidence from before and after to back that up though.

**jkeller** · 10-31-2002, 08:46 AM

Originally posted by alexs
I supplied the IP of the old hamster, the traceroute is the same and the IP goes to the same place where the old hamster was, at Aletia.

IP addresses could not be reassigned so quickly. Hamster was moved just 2 days ago.

Alex

What lookout meant is that Aletia leased servers that were each in different physical locations. As a result, depending on the server you were placed on, you had better or worse access times from your origin. Jaguar's servers are all located in one location, so you'll alwyas get the same results -- no matter which server you're placed on. That can be good or bad, depending on what your originating point is.

I don't think routing tables will do too much because a lot of it has to do with the company Jaguar purchases its conncetions from. If that company doesn't want or isn't able to change routing coming from overseas (remember, a lot of them buy from another, even more limited set of providers from the international level), then Jaguar wouldn't be able to do too much about it.

Each link in the chain is actually composed of its own links (hops), so just changing the endpoints of each link doesn't mean that Jaguar will be able to control what happens internal to each network. I have a lot of hops just to get out of France, and I'm in Paris directly on the France Telecom network (which handles almost all of the international traffic, and out of Paris at that). So ideally I should be in the right place -- but I'm not. I certainly don't think that Jaguar will be able to do anything about that routing. I saw fewer hops to my site when I was at Aletia, but the change to Jaguar has added little to the number of hops/latency, in the big picture of things.

A vote is nice, but unfortuantely it boils down to: are you happy with the services that Jaguar provides? I am, so I'm staying. But they can't offer some of the benefits that this move will provide, including 100% in-house server maintenance, if they redistribute the servers. If that's what you need for your customers, then you might need to consider another solution.

**jason** · 10-31-2002, 09:52 AM

Actually, there is a little bit that can be done to control hops. Generally a data center has connections to several different providers for redundancey--if one goes down the others pick up the slack and the datacenter doesn't go down with the provider. Larger companies have bigger networks, and therefore fewer hops. A provider like Sprint, for example, that has thier own trans-Atlantic lines, will probably have fewer hops than a whole bunch of smaller companies linked together. By adjusting the routing (there are entire college degrees dedicated to this) to use different providers for different destinations they might be able to reduce the hop count. Hop count doesn't necessarilly equal speed, however. Five hops between OC-3's for eample could be faster than a direct 56k connection.

Larger networks accomodate more traffic, but they can also have more backups. Think about a 10-lane freeway in LA vs. the 4-lane NYS Thruway here in upstate NY at rush hour. The LA freeway can hold a lot more traffic, but it still gets backed up. On some of the more rural parts of the Thruway there is much less available "bandwidth," but with less of a demand, there is rarely a problem. I could take the LA freeway directly from Point A to Point B to get home (if I lived in LA) and it could take me two hours. I could take the Thruway to I-390, take 390 to 590, and then 590 to 490 to get to my house in Rochester, travelling the same distance, in a about 15 minutes. I'm going throuh more exits (hops) on narrower roads in Rochester, but the traffic is running faster, so it takes less time.

--Jason

**jkeller** · 10-31-2002, 11:18 AM

Yup, I agree (and understood bandwidth vs. throughput). And I'm certainly not someone with a degree in anything approaching network architecture (I just play one on these forums

).

I didn't mean to give so much time to hops, when latency is definitely a problem (and one that alexs specifically mentioned, natch). What mean is that there is little that Jaguar can do to say that my packet gets routed by Sprint instead of Level3 from France Telecom when they buy a connection from Yipes. If they're a big customer, they might be able to get Yipes to change, which is what I meant by a provider who "doesn't want or isn't able to".

In the end, if alexs is very concerned about latency from Italy for Italian visitors, then he needs to consider a solution that provides that service. I personally think that better service is worth more than a slow or fast loading time -- especially at $100 a year! Speaking from experience, you really pay through the nose in the European professional hosting market.

I know of at least one person who left because the move from Aletia would increase latency. But I think that Jaguar has done the right thing by its customer base as a whole, even if it increases hops. And latency.

**alexs** · 10-31-2002, 11:42 AM

I don't want to go away from Jaguarpc, it was a simple consideration in slower response time after the move to Houston.
I believed in Aletia and now in JaguarPC, and i hope they'll do the best for their customers.
I know they could not change routing tables but they could add another provider for another quality internet connection.

When i choosed Aletia one of the most important things were "services" (now are the same, with Cpanel) and speed/latency (now it's slower).

I doesn't agree about the 100$ a year hosting companies for better response time, before the move i had excellent response time and speed -> at the same price.

Users: Continue to post your vote!

I'm curious about having responses from other locations in the world.

Alex

**JPC-Greg** · 10-31-2002, 01:57 PM

All providers have Sla's for not just uptime but also latency, and from our graphs we arent seeing any latency. If the change in physical location caused your route to run somewhere else (which im sure it did) and somewhere along that path there is latency there isnt much we can do really...but the users hooked directly to that hop would surely be complaining and that provider shouild/would be working on the problem.

We seen some great performance so far with our new facility and as long as Les doesnt walk by and kick the routers (just kidding Les) then we should continue to have this great performance. We're a decent size company but we arent a mega giant, so we may not have the hundred different routes a telecom hub would have in their building....but we do have far more than we need.

For most of the US clients, if not all but those down road from one of the old servers, you will see increased speeds. Houston is a major hub smack in central USA so lots of routes and traffic comes this way. For those in Ca that had to reach an old server, you may have gone through here and then on to NJ which is why its faster foryou now. For those in NJ, well lets face it, it wont be faster than to hit a place in your own city.

All the technical "ms' and ping times aside, you should not notice any decreased speed in your actual sites. The human eye cant see the difference from 50ms and 25ms. If your site is loading slow there is no harm in shooting us a ticket with your trace, your site, and your specific complaint. Maybe the server your on needs a little load balancing or an upgrade of sorts.

Cheers
Greg

**jkeller** · 11-01-2002, 07:52 AM

Originally posted by alexs
I don't want to go away from Jaguarpc, it was a simple consideration in slower response time after the move to Houston.
I believed in Aletia and now in JaguarPC, and i hope they'll do the best for their customers.

Sorry, alexs. I didn't mean to come across as "love 'em or leave 'em". I simply was trying to say that if access times were paramount, then you'd need to consider all the possibilities.

Like I'd posted before, I truly believe that there are factors in any hositng environment that can balance out any changes in hops or latency. Personally, I haven't seen any real differences between my old and new site. And I'd much rather have the on-site support that Jaguar now provides over the co-located servers that Aletia used.

Greg's post, right after yours, gives the best summary I've seen -- and of course, it helps that he knows what he's talking about (instead of me just giving generalities).

You might want to consider taking him up on the offer to open a support ticket with tracert information. It certainly can't hurt!

**Stargazer** · 11-01-2002, 09:12 PM

Hi,

I have several accounts on different servers, and the only one that seems to be slower is Gold. My account on Hampster, now Mercury, seems just as fast as before. Gold has been very slow for me here in NY. My accounts on host24 and host25 were never moved, and show the same performance as Mercury.

Cheers

**lookout** · 11-01-2002, 11:07 PM

Ah, so it's not just me. My traceroutes to Gold are quicker than they were to Dragon, but Gold often seems much slower at putting up pages than Dragon. It's like some monster process is running that really hampers its ability to fulfill data requests consistently. Most pages seem to load in spurts for me, often with at least one part taking an inordinately long time to load. That part is typically internal to the site and different each time the page is loaded. This is very different behavior than what I saw with my site on the old server, Dragon. But it is rather reminiscent of Dragon's predecessor, Puma.

I think my Gold site's visitors from around the world are seeing a significant performance slowdown as well. My daily hit counter is showing about a 33% drop off in visitors. Some of that may be the result of the outages we've experienced this week. But I am also seeing a big drop in the average number of page views per visitor. That suggests problems with the site performance.

**Stargazer** · 11-01-2002, 11:21 PM

This is exactly the behavior I'm seeing. Pages will take a long time to respond at all, then load at a decent rate, with one 800 byte image not loading for up to 10 seconds to complete the page. There have been some times during the day when it's seemed to be ok for a short spell, then it starts slowing again, hanging on one image for 10 seconds or more.

I'm also getting permission errors on the gallery hosted on this account, and have submitted a ticket. I sure hope I see the same performance as my other 3 accounts on different servers in the future. Anyone else have any problems with gallery 1.3.1 on Gold?