This article hopes to shed some light on the misconception of bandwidth speeds and speed testing. More importantly, at the end of the article is information on how to actually discover Internet problems and issues when you are suffering them. Speed testing combined with additional hard data are key to finding and eliminating reliability issues.
Just how realistic are bandwidth test results anyhow?
Have you ever noticed that your Internet is slow and sluggish even when you see good speed test results? At NutPile, we find most public bandwidth testing sites interesting, but overall, not completely useful.
Consumers typically use speed test sites when they want to confirm the amount of bandwidth they are paying for or when they are trying to determine if there are problems with their connections. The results however don’t tell the story.
Update July 20/2016: Here is an article written by someone at CNET who did some research and found interesting results too. Obviously, the services you use to test are also a very important factor.
Speed. Do you need the premium package?
Providers package in creative ways so that consumers feel the need to upgrade to premium packages. For example, you can’t always get static IPs unless you upgrade to the ‘business’ class’. Or, you can’t get unlimited data unless you pay business or certain higher end packages and so on.
When buying Internet access, it’s easy to get roped into long agreements and more expensive bandwidth plans than are needed because the mentality that more is always better often prevails. The reality is that most people could easily downgrade their services and never regret having done so.
For business and, very much depending on location, it may make sense to buy premium. If you live downtown where your provider is very well wired, you’ll probably rarely see issues. If you live a fair distance from downtown or in the suburbs or an area that doesn’t have modern cabling, you might be in for a shock.
No matter which service you have, finding problems will take much more than doing speed tests. Speed tests are useful but it is important to understand the results and what it all means.
Taking the Internet down
The only true method of testing is to continuously send huge files back and forth from one point to another. If large numbers of consumers were doing this all day long it would take the Internet down or a lot more bandwidth would have to be added at countless locations just to sustain such testing.
No one, providers or consumers, can afford the costs of running speed tests nonstop. In addition, if you (and your neighbors) were running these tests continuously, you would be using up all your own bandwidth just testing it, not to mention affecting your neighbors overall services.
Providers count on everyone sharing their bandwidth in order to provide high speed services. Your 50Mbps connection which costs you say $50/Month would actually use up an entire DS3/T3 which costs between $3000 and $12,000 per month depending on location.
In the real world
While testing, there were many times we could barely get to anything on the Internet, experiencing slow, sluggish speeds yet results too often showed we had great bandwidth. Sometimes the results would show less than we should be getting and other times, they showed nearly exactly what the connection speed was supposed to be or even better.
To make sure our testing was fair, we usually ran iperf tests against our servers in various data centers which have very high speed Internet connections to confirm the speed test inaccuracies.
In network testing
A given is that if the test server being used is inside or directly on your provider’s network, the results will almost always be perfect. This is because you are testing from your connection to a server on the same network. Think of it as an extension of your own local network. Anything you connect to inside that network will be very fast because the traffic is never getting out to the Internet.
In the next case, the speed test might try to find the nearest server which could be outside of your provider’s network, or on the edge of its network but either way, optimized to give you the best results.
The less hops there are to that server, the better your speed test results will be. Hops mean the number of routers/switches or devices that your packets have to flow across to get to a location on the Internet.
Even if you pick a test server farther away, it could be (probably?) using an optimized route/network. Only speed test sites could explain this part but based on our knowledge of networking, it seems unreasonable to think that we could download/upload at full speed using up several networks bandwidth along the way.
FYI about hardware and drivers
Factor in what you are testing from. All hardware uses drivers and all hardware has certain limitations. Some hardware is faster than other hardware and some drivers better optimize that hardware. Bad drivers means potential errors and other issues which will not allow the network hardware to reach its full speed. Older hardware or drivers which haven’t been updated or that have no modern updates will not work as well as they could.
Bandwidth vs Throughput
Finally, it is important to keep in mind that bandwidth and throughput are two very different things. Bandwidth is the maximum amount of data or the capacity you pay for that can or is allowed to pass through your connection while throughput is how much data is actually able to travel through your connection which can be limited by many factors.
Practically every service on the Internet has some sort of limiting method in front of their application or their own provider does. Rarely do any services allow anyone 100% access to all of their bandwidth otherwise few people could use the service at the same time as resources would be brought to their knees. Employing various limitations ensures that no one person uses up all of the resources.
Therefore, even if you see 200Mbps results from your speed tests, you will rarely if ever see such transfer rates by typical services other than networks specializing in very high speed downloads.
You share your connection
When you buy your Internet connection, you pick a speed and cost. The speed you pay for is the maximum amount of bandwidth your connection is limited to but the amount of throughput (explained above) is based on sharing with others in your area and a whole lot of other network factors. That sharing by your provider is sometimes called Best Effort delivery.
Providers charge a lot more money for higher speeds that theoretically, you can never really use unless others in your area aren’t using their allotment at the same time.
If you have a number of people at the same location sharing your connection, the more bandwidth you have, the better the chance that everyone will get a good share of it. However, the actual amount of throughput will always be based on many external factors that more than likely prevent you from every getting the full amount of ‘speed’ you are paying for.
How much speed do I need?
Most consumers would be better off picking lower speed packages and saving money because there isn’t much on the Internet that that will let you use that full 100Mbps (12MBs) or 200Mbps (25MBs) speed.
One of the highest bit rates requirements is streaming video. Most video rates from streaming providers are around 3Mbps.
The recommended speed requirements for the new Ultra HD video which NetFlix is now offering is 25Mbps while all others are significantly lower, including HD at it’s highest being 5Mbps.
When you decide on speed, you need only look at what you’ll be using your Internet for and most importantly, how many devices you’ll be using at the same time which have high speed requirements. Most things simply don’t need that much bandwidth.
There are some good ideas out there but none which have been adopted by the masses at this time. What is needed are real ways of testing which would not waste resources while giving consumers a good idea of how their services are performing, speed wise.
In our opinion at least, it remains to be seen whether speed tests are of much value but good or bad, they are yet another tool one can use to get some details when searching for problems and it never hurts to have as many tools as possible.
Solution: Finding problems with ongoing monitoring
A new consumer service called NutPile Networks, gives anyone the ability to monitor their home and business Internet services at little to no cost.
Internet services can be monitored to ensure reasonable levels of service or can be used to gather important connection details, reports and statistics in order to try and get help. Calling support for help without evidence can be a very frustrating experience but being able to gather information should help your provider to look into the issues.
Reports show frequency, length, average times of outages, latency, throughput samples, historical data, optional email notifications and more. A convenient Tickets manager lets you to keep track of each time you’ve contacted support for help. Even know if neighbors or others in the area are down too by sharing graphs.
Simply download the free software, install it on a PC at the location to be monitored and soon enough, things start becoming a little clearer. If a PC is not available PC, a tiny hardware device which contains the same monitoring agent can be purchased and shipped.
Speed testing simply doesn’t show the big picture but ongoing monitoring does. NutPile Networks built the service so that both technical and non technical people can quickly and easily monitor their services.
Please share our site, posts, etc with anyone that could use our services. This small gesture helps us to tell others about our service while keeping costs low.
Try NutPile, no contracts, no obligations, simply sign up and start monitoring now!