A lot of times in this dedicated server forum, I see people comment on the quality of the gigs of transfer they get. I guess some gigs of bandwidth are better than others. Can anyone spell this out for a newbie? I don't know what to look for. I am currently wondering about Webreseller.net's 450 gig of transfer versus Rackspaces's 30 gigs per month. They are not equal are they? Surely 450 is much better than 30. How does the quality come in?
Bradley

well quality of gigs is basically the quality of the lines your provider is using.

For example, if you colo at a datacenter with multiple tier1 backbones (uunet/c&w, etc..) then not only do you have good redundancy but also higher quality bandwidth per say due to the excellent peering and the fact that your bandwidth goes out whichever is the fastest route

Rackshack uses a lot of cogent which doesn't exactly have the best peering, however they are dirt cheap, so it is a tradeoff.

I honestly can't see webhosting being a service that requires multiple tier1 backbones unless you are doing mission critical stuff that requires the fastest speed and 100% uptime.

Unless you are running something that requires 100% uptime and really low pings/latency at all times, rackshack should be fine for you.

It mostly depends on the backbone they utilize. Some backbones just do not have as many peering points as others, and hence less connectivity. Cogent, for example, has alot less peering than Sprint or Worldcom. For this reason you will see many more hops in their traceroutes, bouncing around and trying to keep on their own network until they find a peering point they do have, attempting to stay on their own network for as long as possible, regardless of the hops/routes it takes to do such. However, if price is an issue over quality, then Cogent is the way to go. If you decide Cogent, then just make sure the provider(s) has a backup line, and not running strictly on one provider. And make sure that backup provider has enough capacity at all times to cover the full network load should Cogent go down, or possibly go out of business.

Sean R.
BurstNET

Originally posted by bradley
A lot of times in this dedicated server forum, I see people comment on the quality of the gigs of transfer they get. I guess some gigs of bandwidth are better than others. Can anyone spell this out for a newbie? I don't know what to look for. I am currently wondering about Webreseller.net's 450 gig of transfer versus Rackspaces's 30 gigs per month. They are not equal are they? Surely 450 is much better than 30. How does the quality come in?
Bradley

While Rackspace may have a decent datacenter, Webreseller surely gives you a good combination of quality and price. The Webreseller network is very fast and reliable, maybe even almost as fast as Rackspace. And with their pricing and generous bandwidth, it gives you the best deal for yor money. Rackspace does have a big name in the industry and if you can afford the price and only need 30 GB/month, Rackspace is your best choice.

Quality is a way for a lot of people to justify charging more money if you ask me. If your data gets there, gets there in time, gets there intact, and the links are up 99.99% of the time, then theres nothing to worry about. With webhosting, "in time" for packets is probably around 300ms or less, because frankly, i could put a site here on a 300ms link, and one on a 10ms link to someone, and ask them to load the two, and they probably would never notice the difference, without some tools to *tell* them the difference.

When picking a service provider, ask for download tests, and check uptime history with previous users (if possible) to determine experience, that will probably give you the best overall judgement as some people boast that they have tier-1 bandwidth, but have bad uptime, and vice versa. Also, if you're really concerned about the provider being multihomed, make sure they use bgp4 routing, if they just have one provider with another provider as "backup" thats pretty useless, i've seen a few people with 100mbps from cogent, and a burstable link to 10mbps link from [whomever] to claim they were multihomed, and that was listed as a "backup".

Originally posted by porcupine
[...] i've seen a few people with 100mbps from cogent, and a burstable link to 10mbps link from [whomever] to claim they were multihomed, and that was listed as a "backup".
but you have to agree that it's still better than nothing ...
though everything would be terrible slow.

Rackspace have lower bandwidth allowances because they purchase tier-1 transit from trusted providers, it is more expensive than Yipes/Epik/Cogent etc..., however it is very fast and reliable, and the providers have well built networks and peering relations. Hence Rackspace produce excellent traceroute and ping results and have superb uptime and reliability, this is evident in their reputation and word-of-mouth credibility over the past years. You could easily depend on them.

Webreseller's high bandwidth offerings are relatively new and they primarily use Yipes which is in some serious financial trouble at the moment, I hope they have a backup plan if all goes bad, possibly bring in Cogent or Epik.

Rackspace on the other hand utilize several tier 1 providers, hence if anyone goes bust or offline (for an extended period) then Rackspace won't have a problem continuing to route your traffic and provide the same service as usual. :)

But don´t say "quality of gigs"
Cause Gigs is the unit for data volume, i.e. quantity
Its the quality of the connection
Measured in terms of response time (ping time), packet loss, ...

Quality is a way for a lot of people to justify charging more money if you ask me. If your data gets there, gets there in time, gets there intact, and the links are up 99.99% of the time, then theres nothing to worry about. With webhosting, "in time" for packets is probably around 300ms or less, because frankly, i could put a site here on a 300ms link, and one on a 10ms link to someone, and ask them to load the two, and they probably would never notice the difference, without some tools to *tell* them the difference.

I agree *for example* if you are saying a RS host is *claiming* top quality servers with top quality 'on site' support. Most use the cheap, custom made RS servers and they certainaly do not have their own techs onsite to do server repairs. Look at the complaints here about this. Repairs are restores. Where many issues can be fixed with some simple elbow grease like RackSPACE uses.

However, I disagree, in that most quality assurances are felt over time. Example (using your example as an analogy), I blindfold you and put you in a car. I drive you 100 feet. I then do the same with another car. Can you tell the Yugo from the BMW? Probably not. But over time you will. Same as with a hard drive. Someone uses 5400 IDE and another uses 15,000 SCSI. Can the average user tell the difference? So what you are saying, is they both are the same quality?

Also, I disagree with 300 ms. I can easily tell the diffeence between a site running through 300 ms latency, and one running through, what we average from anywhere in the US, 40 ms.

if you are looking for high uptimes then you must select high quality bandwidth providers like rackspace, dailtone and verio.

Remember quality do not comes cheap. Sure one can provide you with 300GB or 400GB bandwidth but look at the network they are using.

You may feel you are getting good speed from your home towards your server but access to your servers from other countries should also be similar, specially if you consider world wide visitors.

I know many providers who have very good connections from US but very bad or some time no connections from many countries around the world and also from some europe countries which will definitely effect your business and emails.

Bad network often cause in Email bouncing and timeouts. I face such situations many times.

All depends upon your budget and target audience and the amount of downtime and risk factor you can afford.

Quality is a way for a lot of people to justify charging more money if you ask me. That's really funny :)

There is also the factor of how the host's network is set-up.

Some hosts will have multiple great Tier 1 providers but they all run into the same router. Yes, they are running BGP for routing redundancy across multiple lines but if that one router crashes the quality of providers does not really matter. A good host with quality BW with have a fully redundant mesh (along with multiple Tier 1 providers) and we all know that this can be very expensive.