Slow Forum?

[Steve Toy]

Ticket now raised with Bluehost.

[Beechwoods]

Thanks Steve.

[Steve Toy]

An unsatisfactory outcome could see a migration. Steve Powell my guru has already offered to do it.

[MartinT]

If you look at Mark's first screen capture, you will see a list of hostnames going from Mark's computer (with a local IP address of 192.168.0.1), via his adsl provider and various other routers, then via Manchester, London, New York, Washington, Chicago, Denver, Salt Lake City and finally Bluehosts where the AoS is hosted.

The column labelled 'last' shows the last ping time in a series of tests, a ping being an empty packet that is sent to the far end (AoS) to see how long it takes to get there and back. The 'max' column is more interesting because it shows the maximum ping time in the test and you can clearly see that there is a problem from Washington1 onwards as the max ping time is over 0.4s. The average column then shows the average of all the tests and you can see that the majority of the US-based hops are in the red, i.e. over 106ms.

Packet loss is where the ping packet is lost somewhere along the route and never returns. This is a very bad thing and usually indicates an overloaded router along the way, unable to handle all its traffic. Again we can see from Mark's tables that all the packet loss occurs on the US side of the hops.

Nothing can be done about packet loss except to change hosting to one much nearer your target audience, i.e. in the UK or at least Europe. That way you minimise the number of hops that you need to reach the host and the result is that pages are (hopefully) served a lot quicker, provided the hosting servers are efficient and not themselves overloaded.

It's a tricky one but the question that should be asked is whether Bluehosts are giving you an unbeatable price/package for hosting? If not, then you really should consider using hosting a lot nearer your target audience as the routing is poor.

Hope this helps?

[Themis]

Indeed - but can someone explain in layman's terms what the feck "packet loss" is and why lots of it is bad?

The "traceroute" used here, sends a small information and waits for the answer of each intermediate node.
A packet loss, means that an intermediate node (a computer/router between the client and the server) takes too much time to respond. So the packet (small information sent) is considered lost.

This is not the way the forum reacts with the client (internet browser), but rather tests all intermediate nodes to see wheteher they respond on time;

Also, who's to blame for the packet loss - Bluehost or what/whom... What causes it, and how do we remedy it?

I'm afraid all this technical mumbo-jumbo goes WHOOOOSH............ straight over the top of my head!

Marco.

Bluehost is not directly at fault here, there are two possibilities:
A- The routers somewhere in between which are not doing their job. Usually routers/computers fail, but it's the provider (who makes his customers pay for the route use) who is supposed to provide sufficient replacement routers. Usually, the provider is supposed to have more routers than necessary, and to have an automatic switch. Guess what ? they have statistics and try to have as little replacement routers as possible to minimize costs.
B- Part of a backbone is not working: Internet is articulated like a "star" (a "web" to be exact) where there are several "main routes" (big and very fast, called "backbones") and a lot of "intermediate", small connections (usually made by the local internet providers).
When a route is not working, another is chosen automatically. But when a backbone fails, the other routes (and backbones) cannot send all the information and some of it is considered "lost" after a while.

I've tried to explain simply, sorry if it's not clear.

[Marco]

Yes that's much clearer now, thanks (Martin and Dimitri)

I'm afraid that graphs/technical terminology and me don't agree - my (arty) brain just doesn't work that way....

However, I just love the way some folk bandy these terms about as if they were common knowledge!! Ahem....

Anyway, you can rest assured that we will definitely do something about this. Stay tuned!

Marco.

[Marco]

I wonder why things are as slow as a wet fart in the morning, but speed up in the afternoon?

America waking up?

Dunno about others, but everything seems fine now, and the pattern has been the same since the problems began.....

Marco.

[Themis]

Perhaps they have work to do on a backbone, somewhere in the States, and they deactivate it when traffic is minimal : at night.
Perhaps not.

A few years ago, when they upgraded one of the the backbones in Montpellier, it was hell in the evening for one month.

[Beechwoods]

Issues with whatever backbone Bluehosts uses crossed my mind as well. Getting a UK host would minimise the liklihood of these issues impacting our service, but different international hosts will have different backbone routes with different levels of reliability (I'm thinking of those bits of their infrastructure as it gets more local - the European to US backbone offers pretty limited choice). Fingers crossed this is a solveable / temporary problem; if not then moving hosts it will be (and we'll get Tapatalk loaded at the same time )

[MartinT]

May I ask who/how the hosting is paid for? Perhaps a different payment model (e.g. super-members) could help with hosting costs using a better service provider?

I realise this would be a paradigm change for AoS, but I'm just curious and inviting discussion if that's permitted.