Broadband Quality in Sri Lanka: Dialog performs well on download speeds; badly on latency and jitter

Posted on November 2, 2010  /  36 Comments

LIRNEasia‘s recent Broadband QoSE tests report Dialog Telekom’s 1Mbps postpaid broadband package (HSPA) delivers higher download speeds than what was promised in comparison to similar offerings by Mobitel and SLT. However, it does not maintain this trend in measures for latency and jitter. Although there is a difference in underlying technology which most certainly impacts performance, the focus here is on the comparison between what is promised (advertised) versus the actual speeds delivered by the operator, among other important quality indicators.

The latest QoSE report compares three similar broadband packages in Sri Lanka: Dialog Telekom’s postpaid broadband (1Mbps); Mobitel’s Zoom890 (1Mbps); and SLT’s Office (2 Mbps). The results can also be used for regional comparison as these tests were carried out across 11 location in 7 countries.

View results in the QoSE and broadband price benchmark reports available here.


  1. i just created a post containing many speedtests from islandwide.Your point is correct.Dialog’s ping is almost 1 second.But the speed is good.But Airtel is doing a good job i think.Visit this

  2. @ What contention ratio, all test connections are at same contention ratio or different… please comment

    1. We agree that the contention ratio is a critical parameter in delivering performance dimensions. However, what we are measuring is the quality of service EXPERIENCE. What does the customer who pays x or y rupees a month experience? To the customer, the contention ratio is not relevant: she or he simply buys from among the packages on offer. For details on the method please see Professor Timothy Gonsalves’ paper.

  3. Why was this study limited to only these three players? It would be good to get some info about other players in the segment such as Suntel and Airtel.

    1. Dear Aramayel,

      We are a regional organisation. Although this post was highlighting the Sri Lankan results, the tests were carried out in 11 cities across 7 countries in South & South East Asia (2 operators per city, with the exception of Colombo, where we tested packages from 3 operators). These details are available in our report (URL given above). We do not have the resources to test all operators in all the countries we work in; please bear in mind, this was not a Sri Lanka only study. Thanks for your interest.

  4. Why you give a special prominence to Dialog in this study? This report is exceptionally favorable to Dialog with a logo.

    1. LIRNEasia is an indenpendent organisation and we do not favour any operator. We merely presented the results we got from the recent regional broadband tests carried out (please see the report for details – URL above). You may want to pose your question about the use of Dialog’s logo to the Daily Mirror.

  5. They say justice seen is justice done.

    It makes no difference whether you are independent or not, if fail to be seen independent. Wrong choice of words will damage not DM’s, but your reputation.

    I first thought DM report was an advertorial. Please reread and see how it may sound to an ordinary reader. Do you seriously think it is wise to say things like ‘Company X exceeds consumer expectations’? Company X will be very happy, but the competition will feel miserable. You will have more enemies. Is that what you want?

    Wouldn’t it better to use a bit more tactful language than using this marketing lingo? This choice of words serves no purpose unless you purposely want to promote Dialog (and make enemies).

    1. If you read the news release it says, “… the focus of LIRNEasia’s broadband QoSE benchmarking is on the comparison between what is promised or advertised versus the actual speeds delivered by the operators…” – we’re not saying Dialog is better than SLT or Mobitel. We’re saying Dialog performed better than what they (Dialog) promised.

    2. @Puzzled: Think you should direct your obvious anger at Daily Mirror. They’ve just stated facts here. Don’t be too Mad your Mobitel didn’t do badly :)

  6. To err is human, to blame on somebody else is more human. :-)

  7. Why the TRC test results (roughly the same time) is different from yours?

    How come the Dialog, the best according to liren is worst according to TRC?

    Whose word should we believe?

    Who is in whose payroll?

    1. LIRNEasia tested Dialog’s 3G USB Modem package while TRC has tested their fixed Broadband packages (HomeNet and OfficeNet). The underlying technologies of these packages differ, and this has bearing on service quality. Which is why we keep emphasizing – we’re not comparing Operator A to Operator B; we’re comparing what Operator A promised against what Operator A delivered.

  8. I was in Sri Lanka recently (for nearly two months) and was unfortunate enough to get a dialog connection for broadband; a most disgusting service that gave you only a fraction of what was promised. This made me to do some research of my own that led to finding the contrasting reports of the TRC and the LIRNEasia. I was compelled to do my own testing of services with my colleagues at our local counterparts office (they have Broadband connections from 3 service providers) and was convinced that the TRC report depicted a fairly accurate picture whereas the LIRNEasia figures were “cooked up” for reasons best known to them. I think any person could guess the correct answer to Poorna’s question “Who is in whose payroll?” Disgusting!

    1. Our test methodologies and results are transparent and are available on line. As I’ve explained in my response to Poorna, we never said Operator A was better than Operator B – that would be comparing apples with oranges – an unfair comparison. It seems unfair to accuse us of ‘cooking-up’ results when it is clear that Mr. Simons has not referred the test results nor read the articles properly. There seems to have been some misunderstanding.

  9. The possibility can be either way. TRC works more with SLT than with Dialog. What interests me is irrespective of the tester, why it is always the ‘guy we work with’ comes to the top. Is it because we all live in Sri Lanka?

  10. The LIRNEasia office is closed, so it will take a little time to respond.

    Our measurements were taken in October. TRC states that many changes were made in the past few months before it took the measurements. Perhaps we were measuring what existed before the changes and they were measuring after.

    The LIRNEasia method is public and can be used by anyone. Before we comment on TRC’s findings we need to know what specific method they used. Our response will be expedited if the TRC method is made public forthwith.

  11. Just to clarify: TRC appears to have tested Dialog’s WiMAX based fixed wireless packages. LIRNEasia did test these packages back in 2008, but we no longer do because very few people use them. We now only test Dialog’s 3G services using static computers equipped with dongles.

    It’s not very clear what Mr Simons used, but it seems like it was dongle-based 3G (unlikely that one would go through the bother of getting WiMAX for a 2 month stay). How he found TRC results either on WiMAX or on 3G services prior to the past week is a mystery, when the TRC released results ONLY for WiMAX in late December. Even if he had inside information, he would have been comparing apples and oranges: our 3G results and TRC’s WiMAX results.

    To reiterate. LIRNEasia believes that QoS testing is best done by regulatory agencies. We are developing instruments to do the testing and also exerting pressure on regulatory agencies to perform their duty. We have no interest in continuing to test and report once the method is proven and/or the Regulatory Agency does a proper job.

    This is the first time the TRC has released its results. We have been doing so since 2008. We consider the very fact that the TRC is paying attention to this subject a victory. Until we ran ads in newspapers in Nov 2009, they showed no sign of attending to QoS.

    Once the method has been disclosed and is proven satisfactory, we will cease to work on fixed broadband QoS in Sri Lanka and shift resources elsewhere.

  12. There is no Dialog broadband package available at 1Mbps. They advertised the package as “up to 7.2Mbps. I found some of the Dialog towers in Colombo are capable of providing above 5Mbps. But Mobitel’s max is 3.6Mbps. Mobitel say download speed up to 1Mbps. Dialog says up to 7.2Mbps. Can you compare these two??. With this 7.2Mbps, Dialog is under delivered.
    From where did you get this Dialog’s 1Mbps broadband package? Lirneasia has stated “Dialog Telecoms Postpaid broadband (1Mbps)
    So Lirneasia compared Apples with oranges.
    Please correct this.

    1. At the time LIRNEasia tests were performed the tariff table on the Dialog Broadband site had different packages based on bandwidth and download speeds (1 Mbps, 2 Mbps etc.). The ‘up to 7.2 Mbps’ download speed they now advertise, is a recent change on the Dialog Broadband web site. LIRNEasia cannot be responsible for all future tariff plan changes that operators make at their discretion. We will most certainly make the relevant updates to our next round of tests as we always do.

  13. No there wasn’t anything previously at Dialog’s web site to label there service as 1Mbps. Can you prove your point? Dialog’s 1Mbps is your imagination. But it is not right to do an unfair comparison. Others of course, from the very beginning they have clearly advertised their service as “up to 3.6”.
    When it comes to serious matters we can’t work on our imaginations. We should be more responsible. If you have evidence please show us to prove your point. In all occasions Dialog has indicated TRC’s tariff approval numbers along with their packages. As you mentioned we can trace the number to verify where there had been any package In the past for 1Mbps.
    Dialog’s promise is “up to 7.2 and in Mobitel it is “up to 3.6”. Now please compare what is promised and what is delivered.
    At least now you can correct the picture. We would like to see the results.

    1. We do not take screen shots of all the operator web pages at the time of testing (tests are conducted in 11 cities in Asia Pacific). However, prior to replying your previous post I confirmed with Dialog that the text ‘up to 7.2 Mbps*’ under the column ‘Maximum Download Speed’ was a recent (post-October) update and previously the packages were differentiated on various download speeds. Dialog made these changes after TRCs request on publishing maximum speeds. You may verify for yourself.

      Also, just to clarify your previous comment in comparing apples with oranges – LIRNEasia does not compare operators with each other. What we compared was the delivered vs. advertised speeds of each operator/package we tested.

      Our next round of testing will be in October 2011.

  14. Dialog had 1M download – 384k upload package, more retails can be provided

  15. The ‘testing’ done by TRC is a big farce. Neither the tester, nor Anusha Pelpita, having no engineering background, knows anything about QoS testing. All what they have done is to test the pre-established links at TRC. Operators are aware the links are for test purposes. Accordingly, they can configure them to show best performance. Therefore, the ‘test results’ released by TRC have nothing to do with the actual performance.

    I also see TRC has shamelessly copied LIRNEasia’s QoS test methodology with no acknowledgement at all. I do not know whether you guys have patent rights, but if so you should seriously consider legal action. This is openly robbing intellectual property.

    Does TRC have no shame at all?

  16. First of all the forums are interesting to read! but my experience of dialog and airtel broad band packages are really unsatisfactory in-terms of the speed they promise.

    First if we take dialog they promise 7.2 or 1 MBs but their delivery is very much below the line than what they promise.If we take airttel the scenario is worst with addition of connections losses.

    what i see is it differs based on the time the user access Can it be peak or off peak hours when the number of users increases the speeds slows down dramatically and result in user frustration.(I discontinued from dialog and started with airtel now thinking of mobitaal which will be again another bad decision) i am not getting the value for the money i pay this is what needed to be researched from lirneasia.

    Not the mere computational statistic but the speed and the reliability need to be tested with time slots (peak and off peak) and the result should published.Am not saying this report is wrong but i would be much benefited if the above mentioned criteria are been covered.

    The connection i used was the USB dongle unlimited packages from both operators.At some stages i had to get up in midnigt to watch my lectures on the tube.:(

    Hope to see a quality service in the region and i believe these kind of reports will influence toward attaining the quality.Hence the test should done from the users side.

    1. Thanks for your interest. If you look at our results and methodology you’ll see that the testing was carried out on multiple days at six time slots throughout each day to normalise anomalies caused bu peak and off-peak traffic variations. We also report the ‘Value for Money’ aspect on each package tested.

  17. God help Sri Lankan broadband users!!! Any independent person reading the comments so far can see that the last person they are worried about is the customer. By the comments posted, Rohan Samarajiva (RS) and the so called “Transformer” think that broadband testing is rocket science and they have invented it! “TRC has shamelessly copied LIRNEasia’s QoS test methodology with no acknowledgement at all” – what a foolish statement to make! LIRNEasia’s QoS test methodology is nothing but a collection of a few fundamental tests any basic level technician knows to perform. “Until we ran ads in newspapers in Nov 2009, they showed no sign of attending to QoS” – silly! Here again a pathetic statement. RS seems to be only interested in claiming credit for everything and reading through his other comments on different topics on this blog it appears that he is claiming credit for what FCC is doing in US as well! “Transformer” talks about Anush Palpitiya’s lack of Engineering background (I learn that he is the current Telecom Regulator in Sri Lanka). I hope RS has that background!

    And I do not intend to reply to RS’s “It’s not very clear what Mr Simons used, but it seems like it was dongle-based 3G (unlikely that one would go through the bother of getting WiMAX for a 2 month stay). How he found TRC results either on WiMAX or on 3G services prior to the past week is a mystery, when the TRC released results ONLY for WiMAX in late December. Even if he had inside information, he would have been comparing apples and oranges: our 3G results and TRC’s WiMAX results ….etc … etc..” FIRSTLY as my intention was not to get into an argument but to present the actual experience of the customers in spite of what the self-appointed experts have to say and SECONDLY if RS did not get excited (why????) and was patient enough to read my earlier comment carefully, he would have found the answers to all the issues he is raising re the comment including the so called mystery.

    Anyway I have no intention of commenting more on this matter in this blog but will shortly take action to publish (in two major communication journals) our independent findings (which we already have, as stated in my original comment) against the results of LIRNEasia and TRC; together with a peer reviewed critical analysis of LIRNEasia’s incompetent testing method and a discussion on ethics and sincerity expected of research organizations like LIRNEasia.

    1. We look forward to comments on how our method may be improved. Publication of the TRC method would also give us good ideas and would help all the stakeholders. Please do keep us posted on the publications. Thanks for your interest.

  18. @shazna,

    /We do not take screen shots of all the operator web pages at the time of testing (tests are conducted in 11 cities in Asia Pacific)./

    You make this statement as if taking a screen shot is such a time consuming task. For your information, it is just Alt+Printscreen and then Ctrl+v in MS Word. Save it and you are done. Less than 10 seconds. May be one minute, even for a blonde.

    When you place emphasis on advertised speeds it is your responsibility to have the necessary information to support your claim, in case you are challenged at a court by an operator. I wish I see you claiming not having taken screen shots as you did testing in eleven cities, under cross questioning. If I were the prosecuting attorney, you can guess my next question!

    Now for not having done that 10 seconds task, how many minutes you waste in phone calls?

    Please do proper research. Don’t do half baked research and use pitiable excuses. It is worse than doing nothing.

    1. We intend on taking screen shots during our future tests. We cannot do anything about the past. We do however, have confirmation from Dialog that package details on the website were updated recently.

  19. /“Until we ran ads in newspapers in Nov 2009, they showed no sign of attending to QoS”/

    This is the understatement of the year, not to mention that is comes for a former DG. TRC’s broadband QoS work started much earlier, in fact under the previous DG though they didn’t want to rush in releasing results.

    Mr. Sanath Siriwardena, who is in charge of the test process at TRC was involved in broadband testing for at least five years and has widely published on the subject.

  20. Hello Mr. “Ted Simons”,

    What hilarious comments you make!

    Why you do not say a word about TRC testing lines selected by the operators themselves? This is like training a Montessori kid to welcome the chief guest at an annual concert. Everybody knows the best kid gets the opportunity and the school’s performance should not be measured by his speech. TRC should be ashamed to do this kind of gimmicks and call it testing. Is that what you call quality testing? Do SLS officers ask factory owners themselves to select the test material? Anusha Pelpita perhaps with his zero knowledge in engineering and telecommunication might think so but not others.

    QoS testing is not new but we have witnessed LIRNEasia doing it in a certain style. For example, they did not use any of the commonly available ping based test tools available in the net. (which they could.) Instead they develop a software application to measure actual time to download a file of given size. This is what they introduced new and TRC shamelessly copied with no acknowledgements. You have also forgotten that TRC also follows the same presentation style followed by LIRNEasia. So what is new in TRC testing other than doing it on lines selected by operators? Haven’t you copied the methodology from LIRNEasia? Or do you say Anusha Pelpita invented it?

    You think you are smart in coping thing but you are not. LIRNEasia tests from 8 am to 11 pm. You have added one hour to both and test from 9 am to 12 midnight. Very smart no?

    Whatever said and done LIRNEasia has made its application and methodology public so that anybody can test for themselves if they are not satisfied with the results? Why not Anusha Pelpita too does the same? If Anusha Pelpita can make his application public I challenge that I test an ordinary line and prove his testing is nothing but a big farce, done only to appease King Mahinda. Pelpita is only trying to earn the credibility he lost when Defence ministry published laws restricting multiple SIM use even without TRC DG aware about it. He sure is a very competent DG I think. That is why others take the decisions he should take even without informing him? Doesn’t he still feel ashamed?

  21. Machan, transformer, whatever said and done, I don’t think TRC employs blondes to do QoS testing. :-)