Tuesday, May 31, 2011

With broadband speed you don't necessarily know what you will be getting

Since I live in an urban area, I am luck enough to now have a choice of Internet service providers vying for my business. Competition is definitely a good thing from the customer's point of view, but one of the problems is that many consumers don't know if they can believe the claims of the competing service providers about the speed of their Internet services.

Recently, I was delighted to receive the following text from my broadband provider: "UPC has increased your broadband speed from 15Mb to 20Mb Unplug your modem from the power supply, plug it back in and you'll have even faster broadband. Enjoy." Clearly, this was good news, but since I am sceptical I decided to use a SpeedTest.net testing service to test my Internet service both before and after the upgrade to see if the speed I experience really was as high as claimed.

If you read to the end, you can see the full details of the testing I did of my connection speed before and after the upgrade, but if you want a quick summary here are the conclusions:
  • The download speed achieved on my download link did indeed increase as promised by my ISP, however in most cases it would not have any noticeable effect upon the overall experience of using the Internet because download speeds from many sites was way slower than the maximum capacity of the link from my ISP.  Therefore, although any increase is welcome, it is probably not worth my while to spend more money on a further increase in download speed.
  • The upload speed of my connection was much slower than the download speed. This did not change noticeably after the upgrade which is a pity because this would be a factor that would impact my experience of the Internet. Upload speed should become an important factor when comparing different Internet providers, but very few publicize their upload speed.
  • Some of the sites which were giving me the fastest download speed also had considerable ping delays. In some applications (e.g. VOIP calls) the ping delays I measured would be likely to have a noticeable impact upon performance. 
When considering the speed of your internet connection, one of the first things that you need to realize is that the internet is not a single unified entity. The speed of the connection from your house to one particular server on the internet might be quite slow, but your connection to other servers might be fast. In order to account for this, the Speedtest site has a wide range of test servers all over the world and you can choose which one you can use for testing your speed.

Before I present the actual test results, I think I should briefly explain the parameters that SpeedTest measures about the link. It measures three factors:
  1. Download Speed: This is metric shows how quickly it is possible to download a large file from the test site. It is measured by downloading test files of various sizes and then measuring how the time to download varies with the size of the file i.e. they ignore the overhead delay which would be required to start any download. This is the metric most commonly quoted by ISPs in their advertisements because one of the most common uses of the Internet when it first became popular was to download large software kits. However, the things that people do on the Internet has changed significantly in recent years so this is not necessarily the most relevant metric for current users of popular sites like Facebook etc..
  2. Upload Speed: This is a measure of how quickly you can upload a large file to the internet. It is essentially the same measurement as the download speed except that the data is going in the reverse direction. This would be the metric that would affect how quickly you could upload some pictures to your favorite photo sharing service. Many Internet links provide significantly less upload speed than the download speed because they assume that you will spend the bulk of your time accessing information that other people have uploaded rather than uploading your own information. However, since many people spend a lot of their time online using social network sites it is not clear to me that this assumption is still valid.
  3. Ping Time: This is a simple measure of how long a single packet of information takes to travel from your computer to the chosen Internet server and back again. This measure is important in any application which involves real-time interaction over the network because this will set the minimum on the communications lag. For example, many people will have seen remote correspondents being interviewed on the TV news where it is clear that their is a noticeable lag between the interviewer asking a question and the question being heard by the interviewee. As you know packets are not sent directly from your computer to the Internet server, instead they pass through several intermediate nodes with each hop adding to the delay in delivery time.
These are not the only factors that need to be measured to judge the quality of the line. Other factors include the percentage of packets lost and the jitter (i.e. how much does the ping time vary from one packet to another), but to make my life easy I only considered the parameters that Speedtest gave me.

I ran the first test utility before I rebooted my network router so this would be a measure of the performance of my existing network link. By default, the site assigned me to use a server in Limerick for testing, but the performance of my link to this server was surprisingly poor so I repeated the tests with 3 other servers in more distant parts of the planet. I took each measurement twice to get a quick feel about how repeatable the test result might be.


By default, the speedtest site selected a server in Limerick, Ireland to test my connection. Although Limerick is not very far from my home (roughly 200km) I found that the results from this server were quite poor (perhaps the server is overloaded or else it might be in a data center without fast connections to other parts of the Internet). I chose a few other servers in different cities to repeat my test and as you can see below the tests varied quite a bit depending upon which server I selected.


The speed of my link to Oslo and San Jose was reasonably good, but the link to Philadelphia was not as good and the link to Limerick was worst of all.  It is surprising that it took longer for my network packets to reach Limerick (which is roughly 200 km from my house in Dublin) than they took to reach San Jose (which is over 8,000 km away) - but of course the packets destined for Limerick could very well be routed through intermediate servers on the far side of the Atlantic.

Download Speed (Mb/s)Upload Speed(Mb/s)Ping Time (ms)Server Location
0.97/1.970.94/1.03215/220Limerick
10.27/9.230.99/0.95115/120Oslo
2.25/1.991.08/0.99123/124Philadephia
9.07/6.750.84/0.82176/182San Jose

I repeated the same test immediately after I rebooted the network to see if there was a noticeable difference. When I ran the pre-upgrade tests I did not realize that there was a test server available in Dublin. I ran tests with this server and I was pleased to see that the speed achieved to this server was over 15 Mb/s so the link from my house must indeed have been upgraded beyond 15Mb/s. The ping time to the server in Dublin was dramatically better than any of the others which probably means that there were very few networks hops between my home and this server.

Download Speed (Mb/s)Upload Speed(Mb/s)Ping Time (ms)Server Location
0.98/1.390.80/0.84217/231Limerick
10.38/11.181.05/1.03122/110Oslo
1.70/2.101.15/1.06121/124Philadelphia
10.18/10.881.23/1.10178/178San Jose
15.98/15.360.94/1.0231/30Dublin

In general there was not a noticeable difference in the speed of the link to any of the servers I had used before. Of course the speed of an overall network link is determined by the speed of the slowest link along the path. Looking at download speed, there is a dramatic difference between the speed when using different sites. This seems to suggest that the so the link from my house to the ISP's data center is not the bottleneck and the bottleneck must be somewhere else along the path to the slower sites. Therefore speeding up the capability of the link between my house and the ISP is not likely to affect the overall download speed achieved from most of the hosts.

However, the upload speed tells a very different story. The upload speed was around 1 Mb/s for any of the servers tested which is much less than the download speed. This means that the upload speed was probably limited by the link between my house and the ISP's data center.

Using a VOIP tool like Skype, the same amount of data would be transferred in both directions. In most cases it is the upload speed of each parties Internet link rather than the download speed that would determine the overall quality of the connection achieved. Therefore someone who is unhappy with the quality of their Skype link would be wasting their time and money in switching to an alternative service provider who promised higher download speeds.

Of course network performance will vary depending upon congestion so I did the following tests the next morning. It seems that the download speed from Limerick had improved from the day before, but the download speed from the server in Dublin was worse. The upload speed was still remarkably similar for both servers although it was about 20% faster than the night before.

Download Speed (Mb/s)Upload Speed(Mb/s)Ping Time (ms)Server Location
3.05/2.931.27/1.1.6216/214Limerick
12.93/12.721.29/1.3327/24Dublin

This asymmetry between the upload and download speeds is mainly a feature of the domestic broadband market. Commercial services are more likely to provide matched upload and download speeds. To see how a commercial service would perform I repeated the tests from the office where I work.

Although I the office is physically in Dublin, my employer has an internet connection provided by a UK based company. Therefore the SpeedTest.net service defaulted to using a server in London which it was able to reach faster than the one in Dublin. Luckily my employers has a very good internet link, but the download speed was 14-15Mb/s which is only slightly better than the speed I was getting at home. Looking at the upload speed however, the connection in the office was about 10 times faster than my home link

Download Speed (Mb/s)Upload Speed(Mb/s)Ping Time (ms)Server Location
14.38/14.2312.90/13.8256/77Dublin
15.10/14.0910.68/13.1024/24London

I mentioned before that part of the reason limiting the speedtest results are the fact that my packets were competing with packets from other network users (this is called network contention). Therefore I decided to test the speed of the link again early on a Sunday morning when I guessed that there would be relatively little traffic congestion on the network. As you can see from the results below the download speed was indeed much faster, but the upload speed was virtually unchanged.

Download Speed (Mb/s)Upload Speed(Mb/s)Ping Time (ms)Server Location
19.52/19.161.25/1.2426/25Dublin
2.75/2.101.26/1.28215/215Limerick
12.38/12.901.37/1.38110/109Oslo

Saturday, May 28, 2011

Could I run a Triathlon?

I like to keep myself fit and healthy through a reasonable amount of regular exercise. However, I have mixed feelings about the idea of running a triathlon. On the one hand I think the athletes are foolish to subject their bodies to the strain of such an event, but I must also admit that I am envious of how fit they must be to be able to complete such a tough course without collapsing.

A colleague recently told me that he was entering into the Tri Athy event. This is billed as "The Perfect Introduction to Triathlon" because the relatively short course that involves a 400m Swim followed by a 13.6km Cycle and a 3.6km Run. I did give some consideration to joining him for the event, but then I chickened out at the end because I was not sure if I would be able for even this shortened version of a triathlon.

This morning I was visiting the gym and so I decided to do my own version of a mini triathlon to see how I would get on.
  1. I started by swiming 20 lengths of the pool which is equivalent to 500m. This took me about 17 minutes which I think is a reasonable pace. I swam crawl all the way and didn't stop at all, although I slowed down quite a bit nearing the end.
  2. Next I went for a cycle of about about 15km which took me slightly over 4o minutes. I was trying to be as close to triathlon conditions as possible, but I didn't run out of the gym onto my bike in my swimming gear. Instead I tried off properly and changed into my cycling gear. If it was a real triathlon, I think that the transition from swimming to cycling would be hard to do properly and I would certainly not relish the prospect of cycling in wet gear in Irish weather.
  3. I was feeling Next I ran a 6km jog around a nearby park which took me 36 minutes. I was wondering how I would be sure how far I was running. I used the My Tracks application on my phone to monitor my progress and a new version of the application was pushed out just this week. I was pleasantly surprised that with the new version a robotic voice informed me as I passed each kilometer marker and it gave me statistics about my running pace so I never had to look down ad the screen to monitor my progress.
At the end of all this I was quite tired, but I am happy to report that I was not close to collapsing. Perhaps with a bit more practice, I will feel able to tackle the introductory triathlon next year.

Friday, May 20, 2011

Do we really need competition in the electricity market?

Ireland has only relatively recently introduced competition into the residential electricity market. As a result a considerable amount of broadcast time on both radio and TV is devoted to advertisements from the various electricity companies encouraging household to switch electricity suppliers. Naturally, I took advantage of this situation to switch to a company who is providing me with cheapest electricity, but at the same time I am confused by what is the point of the exercise.



I am not an economist, but as far as I understand the reason why they recommend the introduction of competition into markets is because the competitive pressure to win market share will encourage the various companies involved to introduce innovations that will reduce costs and/or increase quality of service.



I can see how this works in the telecoms market. Some of the companies active in the market provide high quality services and premium prices and others provide lower levels of service, but at cheaper prices. In practice it can be difficult to find out the quailty level of the service you will get from a particular service provider (in fact it can sometimes be hard to find out the exact price they will charge you either), but in general the market works well with different service offerings appealing to different market segments.



The trouble with the electricity market is that the electric grid is designed to deliver exactly the same voltage of electricity to all of the customers. I know that some of the companies like to boast about the fact that they use a high percentage of renewable energy sources such as wind, but there is no way to tell that the electricity delivered to their customers' houses is from these clean source sand not from the dirty coal fired stations owned by their competitors. Likewise, if the wind didn't blow and there was a shortage of electricity, fairness should dictate that the environmentally conscious customers should be the ones subjected to the inconvenience of a temporary loss of supply - but there is no way to cut off supply to these customers while maintaining supply to the customers who chose to go with the less environmentally friendly company.



As far as I can see the only way that the electricity companies can compete with each other is in their billing system. This is hardly the area where the greatest amount of innovation is to be found. Am I missing something?

Friday, May 13, 2011

The importance of regular excercise

I am a big fan of getting regular exercise. However, instead of being intensively involved in a single sport I normally try to fit exercise into my normal daily routine. I normally only use Ben Hardill's wonderful Tracks2Miles application to upload my DailyMile records when I undertake a significant run/cycle. However, most of my exercise is accounted for by the fact that cycle to and from work when the weather allows. I also try to fit in a short walk at lunch time each day - I find that the fresh air helps improve my productivity in the afternoon.

It is surprising how few IBM employees at the site where I am based realise that there is a beautiful walk around the perimeter. The complete circuit is 4.5km, but several shorter options are also possible if time is limited. Although parts of the path can get a little bit mucky during the winter months it is really beautiful. Trees have been planted that obscure the view of the IBM buildings and at times you would find it hard to see the IBM buildings and so when you see the cows on the neighbouring farms, you would find it hard to believe that over a thousand people are working in high-tech jobs only meters from where you are walking in the middle of nature. It is very common to see hares sharing the track and even the occasional fox (probably looking to feast on the Hares).

View Lunch walk Around IBM campus in a larger map

Incidentally I am no longer mayor of my local gym. I have not been back since I was ousted from the mayorship at the end of March. It is not that I have been sulking it is just that it seems a shame to be indoors running on a treadmill when we have been experiencing a rare extended spell of sunny spring weather in Ireland.

Monday, May 9, 2011

My new job is offically announced

When people ask me what I do for a living I normally tell them that I am a software engineer. This is not entirely a lie, because I do have an engineering degree and I work in a software engineering related job. However, for over a decade my job has essentially been that of a manager. While I am proud to say that I have been able to keep up with the latest trends and technologies in software development - a very small proportion of my working days were spent writing software and a large percentage of my working days were spent doing general management tasks.

I have always believed that the work of the engineers in a high-tech company is much more valuable than the work of their managers. However, in recent times I had direct or indirect people management responsibility for over 100 employees. The vast majority of these employees were able to do their job with minimal input from me, but when you have responsibility for such a large team it is inevitable that you end up having to spend a lot of time 'managing' and hence have very little time left for 'engineering'.

Towards the end of last year I had career planing discussions with both my mentor and my own manager. It became clear to me that if I wanted to be successful while staying in the same career path I would have to devote even more energy to the management side of my role and hence reduce even further the limited amount of time I was spending on direct engineering work. This is not what I really wanted to do with my life and hence I decided that I would attempt to change my career path away from management and towards engineering.

It took a few months for me to find a suitable new role and to also arrange a replacement for my management position. However, I am delighted to say that the move has finally been formally announced and I am starting my new career this month.

I will still be part of the Sametime development team, but my new job responsibility will be to lead an imitative to simplify the way the product is installed and configured. Many of our customers have been telling us that they love the wonderful new capabilities of the most recent versions of Sametime, however they are less thrilled with the complex steps required to get the product installed and configured in a way that is compatible with their network infrastructure. It is easy to recognize that this is a problem we need to tackle, but it will not be easy to solve it.

One of the great things about working for a company like IBM is that it is possible to rise to a very senior level without having to be a manager. However, the more senior a person is, the harder it is for them to make such a dramatic career change. Because I am reasonably senior in IBM, it will be hard for me to make the change successfully, and still meet the expectations of someone at my pay grade, but if I postpone the move any longer it will only become more difficult. I have no doubt that it will be very difficult for me to become a high performer in my new role, but I am looking forward to the challenge and I know deep down that I am making the right move.

Update: I posted more information about my new role on my IBM internal blog.

Friday, May 6, 2011

The importance of play in successful organisations

As my friends know I am a great believer in the importance of taking time to ensure that I and my colleagues have fun at work. In fact "Have Fun at Work" was one of the official corporate values at Lotus when I first started in my current job. Unfortunately IBM does not have a similar corporate value, so it lapsed as partof the  take-over.

However, many IBMers do some great work to ensure that IBM is a really fun place to work. A large portion of the key leaders at the Dublin Software Lab are former Lotus people and as a result of this many of our employee surveys speak highly of how it is a really fun place to work. I think we need to be careful that we don't lose this culture because it is a valuable recruiting an retention tool.

I recently found this wonderful TED talk by a Primatologist named Isabel Behncke Izquierdo where she explains much more eloquently than I ever could why the human race could increase tolerance and harmony by learning from studies of how Bonobo society values play.

Tuesday, May 3, 2011

"Spousonomics" is not a word I plan to use

Some people use their blog to write reviews of books that they have read. However I am going to use mine to write about a book that I have no intention of ever reading.

When I visit my local library, I normally visit their "new arrivals" shelf to see if there are any interesting books. On my most recent visit I spotted an interesting looking book cover with the catchy title of Spousonomics. From reading the blurb the book seems to be about how you can use lessons from economic theories to achieve a happy marriage. Not surprisingly the authors are American and have a web site to go with the book complete with a Facebook page, a Twitter feed etc.

I can't be too critics of Americans because technically I am a US citizen - but one of the most laughable aspects of American society is a belief that all aspects of life are explainable by economics. I find this book title hilarious but, I have no intention of reading it and I think that it would probably be detrimental to my marriage to start quoting this rubbish.