By Peter Cochrane, 3 September 2008 13:14
COMMENT
Initial vibes received in Jamestown, California. More comments picked up in Cupertino. Dispatched to silicon.com from home two weeks later after a bout of pneumonia that downed me for more than 10 days
These days home and office users are able to order and install their broadband connections unaided. And for the most part they have performed as advertised - if you read the small print.
What does 'up to 8Mbps' actually get you? Mostly a slow dribble for download, and even worse, five per cent or so of that for uploads. This was inadequate yesterday, and will kill web 2.0 economies tomorrow.
It's not surprising, then, that we have seen a rapid rise in public complaints as well as political and regulatory activity in the past couple of years. But at last the industry is responding with a few real programmes involving upgrades heading toward a genuine 50Mbps service.
So real broadband is coming? Perhaps and perhaps not. Now there is a new problem looming that puts the onus on the end user instead of the supplier. The reality is that eventually a lot of users will buy and be supplied with 50Mbps but they will only be able to access 25Mbps. How come? Poor virus protection software and old hardware are the first impediments.
I first became aware of this in the US where companies such as Verizon are delivering real broadband to the home and office. And just a week ago I was privy to observing the broadband speeds of a variety of laptops running different antivirus applications.
The chart below shows the per cent data throughput on a modern, high performance machine operating at two extremes, with the best and worst antivirus software identified on the market today. Notice the cliff-like fall off in performance at 10 and 25Mbps.
The second series of tests shows a comparison between the modern machine and a three-year-old model. The individual antivirus products tested are labelled 'a' through 'g'. This presents a far more drastic set of performance challenges.
Two more factors that could hamper broadband speeds are firewall software and operating systems. To date I have no hard evidence but I have had conversations confirming that there may be a few 'elephant pits' waiting here too!
It seems to me this class of problem will open up the need for honest online testing and diagnostic services. And broadband customers will have to be sure it is not their hardware and software that is the problem before they start complaining to their provider.
Surely the telco/netco/ISP quarter will also use this information to defend their delivery against contracts that suggest huge speed increases. They may be delivering but the customer may not be seeing it!
Comments
There are 12 comments. Join the discussion
1. DWL
Obviously, the AV/Firewall software is executed on the same general purpose processor that executes the operating system.
One way of increasing throughput would be be to encapsulate the 'Dataflow Dominated' tasks of the firewall such as deep packet inspection within an integrated circuit, and so offload those tasks to dedicated, fast, power efficient hardware rather than slow down the general purpose processor.
In this way, security would be a hardware feature integral to the modem (like modulation and coding are), rather than a bolt-on to detect problems once they are hiding in the operating system.
2. anonymous
Hallo Peter,
As usual a very interesting and needful article. What would make it even more interesting is to publish the comparison naming the AV software, but I suppose that is not going to happen! However what is happening is that, by its very nature, AV software also impacts on the speed of the computer and that comparison would be very worthwhile. I notice a major difference between different AV programs and I suspect the slower ones would also prove to slow down the broadband the most.
Regards
Keith
3. Simon
While the points you make are valid, this still ignores the most fundamental issue - that services are being sold on the basis of what the end users ADSL circuit will carry, and not what the whole circuit back to the ISP is capable of.
I recall when I got my first ADSL, literally within days of my exchange being enabled. Back then the services were marketed on speed and contention ratio - so I knew that in the worst case I was only guaranteed 10kbps download.
I think we'd see a lot less complaints if (like some other countries) our circuits were marketed on committed rates. I'm sure it would change the market somewhat if people were marketing "up to 8mbps, 100kbps guaranteed" which seems to be about what some ISPs are offering.
4. Mike McNamara
Interesting figures, just shows how much more work many software vendors have got to do to keep up with ever increasing broadband speed (but sadly not in the UK for the time being).
Are you able to name the Antivirus software vendors that were used in this test?
I keep a track of my BB speed using a cross-section of online speed tests to get a good average figure. I think I'm lucky to stay around the 4MB mark, but of course nowhere near the 'up to 8MB' con that still gets promoted.
5. Peter Cochrane
DWL = Yep there are solutions - but not out in the general market. And it will take a while....meantime people will complain! Peter
6. Peter Cochrane
Simon = My (perhaps to) oblique point was that the delivery by misrepresentation of reality will have to stop. ISP/NETCOs will have to come clean. THE AV software comes into play to confound us! Peter
7. Peter Cochrane
Anonymous = I never do product comparisons - but I'm sure someone will - it is a regular feature of the trade mags. Peter
8. Alastair Macfadyen
A friend of mine pointed out to me, in the course of some discussions about problems with Wi-Fi connectivity, that having the link encrypted (WEP or WPA) also has a disastrous effect on the data rate. I tried this on my wife's laptop and found that I was getting about 3MBps on a supposed 54Mbps link, in other words only about half the expected rate.
From your remarks it's possible that some of this is down to the Firewall, though.
I see that you use Wi-Fi a lot on your travels, what is your experience in this area?
9. Peter Cochrane
Alastair = In truth I don't use much coding or encryption with any of my work. But when I do I always pay a price. Unfortunately there is no free lunch, but your loss level seems excessive to me.
My global experience of wifi varies vastly.
For the most part it is good, but occasionally I get a really bad link - and can never decide if it is the backhaul land line rate, ISP congestion, or whatever!
Peter
10. Chris Mead
Fascinating stats - but does beg the question are all AV packages equally effective? Presumably not - but is there any correlation between AV effectiveness and interference with broadband performance?
11. Craig
Excellent! Those of us who don't run AV software (using Mac OS, haven't used AV software - ever!) can look forward to ultra-fast broadband without any reason not to receive the fastest possible connection! Somehow I think there will be *another* reason my connectiom is running slow I'm sure.
12. Andrew Went
2 points need to be made here:
1. The variable rather than committed rate is based on a number of factors being tested on the line by exchange equipment once the service is up. It cannot be done until the service is running. It also involves consumer choices between low latency and high throughput. Gamers often sacrifice their download speed for better latency.
2. The guy with the Mac: you are very lucky, a very light user, or blissfully ignorant of all the baaad stuff on your Mac. They are viruses, worms and other forms of malware written for Mac. I snagged a free G4 recently that was barely running. It had no cookies or personal files remaining. After a cleanup with a proprietary firewall which found nearly 100 infections it was running like new again...