Internet: when delivery falls short of promises

Our in-house manual contains detailed instructions to each of our offices on how to deal with internet problems including all local steps to be taken before calling tech support. In Malaysia, however, it contains some additional items: we have been through the complaints routine with the Streamyx can't-help-desk so often that we even know their script, their responses to certain replies and how to tailor our responses so as to avoid them announcing that it's all our fault and putting the phone down.

The tech support desk is a call centre staffed with dumbos who know nothing about the tech or the service save what is in their script. Mention Linux and suddenly you become non-person, very politely told to have a nice day (as if that's possible when the tech's broken again) as the call is terminated.

Let's give you an idea just how bad this is: the service is on a good day slow; on a bad day it doesn't work at all; on really awful days, like the past two days, it works sometimes, for a little while. During a one hour period this morning, it worked for inbound but not outbound mail; then it worked for outbound but not inbound mail; some websites worked, some didn't. We could access some of our own servers and not others even those within the same country. Our VoIP started to fade in and out, sometimes outbound sound was inaudible and other times inbound sound was silent. Then that, and Skype, which had worked a bit, went off all together - as did everything else.

This isn't a problem with our tech: we have, in the office where there is currently this problem, three different operating systems and five different browsers, three different firewalls on individual PCs. They all go through one router (hard wired) and the same "modem." The router and modem both report that we are properly connected to the internet. The internet doesn't agree.

The problem is clearly somewhere the other side of the ISP's authentication server. But the process of making a report - which we do with ridiculous frequency to the point where Streamyx parent company Telecom Malaysia gave us a refund on our phone bill because of the amount of time we spend talking to Streamyx - requires them to determine whether the problem is here. It isn't: it never is, unless you count the time that we told them that their modem was defective and they should replace it but they didn't want to do that: when they did, our connection speed increased by 20% to achieve the heady heights of 700kb/s - against a contracted 1000.

So we've established that the system is rubbish and the support desk staff are clueless and the system for dealing with complaints seems to be to do nothing and hope the problem resolves itself. Which, of course, it doesn't but people just get used to dismal communications.

So do we say that Streamyx and its government-owned parent are useless? Actually, no. Streamyx and TM are at the forefront of a revolution that other developing and transitional countries are going to have to go through, and it's very painful. And it's all to do with economies of scale and cost. And the reality is that they do the best job they can in very difficult circumstances.

First, Malaysia has a particular problem with demand: the vast majority of its users communicate largely or even exclusively with websites and servers outside the country: this is because they are seeking content in English. This is the opposite of, say, Japan and Korea where local languages mean that most users visit local websites.

This leads to a massive demand for capacity for overseas cables - but TM has only one major overseas cable and it suffers from both overload and breaks.

Secondly, a standard 1MB account costs as much in Malaysia as it does in the UK, but in Malaysia, people earn around a fifth of UK salaries. Also, the size of the market is tiny: the population is only 1/3 the size of the UK but the number of households is smaller because family units are bigger and people live at home longer. And much of the population lives far away from a telephone exchange capable of supporting the higher demands of broadband, and the cost of getting it to them is high. So Streamyx is constantly struggling for investment capital. And the position is made worse not better by the attempts to increase broadband takeup by introducing limited use but very cheap connections - as little as the equivalent of GBP3 per month. But the set up costs are the same as for the more expensive services and those who take up such an offer are likely to be experimenting with the technology and therefore likely to be even heavier downloaders of e.g. MP3s.

It is true that Streamyx does invest in things that some might see as peripheral - such as its BlueHyppo service but in the great scheme of things, if they are paying their staff market pay and conditions instead of government sector terms, and if they are performing to market performance standards, it's not a costly service to run. But where money could be saved is in advertising: the truth is that the system simply does not deliver the performance that its customers have a right to expect.

Aside from an overhaul of its dismal helpdesk, what can Streamyx do to fix the problems?

First, it has to concentrate on getting the service right for existing users: many are already looking at alternatives including, as are we, a satellite connection run out of Hong Kong. The first thing is to improve the complaints handling procedure in some of the most basic ways. The second is to purchase more and faster capacity. Even a doubling of existing international capacity would not, we think, bring performance up to an acceptable standard. And our problems in the past few days have been a recurrence of an old problem where the server that resolves address requests isn't working properly - even websites within Malaysia could not be reached. Twenty four hours with a service that might lose one's work at any moment is not acceptable - but it's where we are right now.

It has to start to communicate with its customer base. This notice is on the website today: tm net sdn bhd would like to inform its customers that there was a power outage at its data center from 1:50 p.m. until 2:11 p.m, today, 2nd october 2006. It's two days old. And right at this very moment, we are getting the early warning signs that it's all about to go off again: some machines cannot get to some websites.

Well, it didn't go off completely: but it cost ten minutes while we decided whether it was safe to continue to work.

TM or other providers need to cable apartment blocks: shared twisted pair wiring rising, in the case of our Editor's flat, some 400 feet up is not man enough for the job.

Malaysia just happens to be at the point where it needs the techology that the developed world has but it can't achieve the critical mass at the necessary prices to afford to put in a world class system, nor anything like it. Singapore next door has about a quarter of the population, with almost no "rural poor" in a land mass the size of metropolitan Kuala Lumpur. It's no wonder that Singapore has wired more than 90% of its population with superfast cable at very affordable prices.

We recently opened an office in Labuan. After four months, we gave up and closed it, the primary reason being that the broadband connection was so unreliable that the office worked an average of less than half a day daily. Internal communications were next to impossible except by phone.

There are premium routes to address the problems in KL: some office blocks do have high speed reliable internet connections. But for distributed organisations, such approaches do not work. Most of our people work from home or from small satellite offices with just two or three people in them, generally close to where concentrations of staff live so that there is no need to commute: in the case of one office, for example, all four members of staff live within a three minute walk of their homes. So from the point of view of an ISP we are just host of small account holders, even though there are multiple accounts in the same name, in several locations across the country.

The question is, ultimately, a political (small p) question: is the priority to make fast, bulletproof lines available in city centres, increase revenues from those and then expand to more marginal districts or to try to make blanket coverage early in the process, knowing that the service will not meet the expectations of its more sophisticated customers.

It's a difficult balancing act. TM do quite well given the objectives and constraints: but being generous enough to recognise that doesn't mean that we don't occasionally want to find someone at TM and give him a damned good shaking. Politicians are beginning to complain and the internet bulletin boards are awash with some very blunt comments about the service. It's difficult to sit in one of the world's most modern cities with a skyline that may be from a science fiction movie only to find out that something as simple as comms has been left behind; and it's even more galling when one realises that seven years ago, Malaysia installed what was then, according to some reports at the time, the most advanced and fastest internet service available anywhere (disputed but only in relation to timing, not as to the actual tech).