|
||||||||||||||||||||||||||
|
|
INTERNET ACCESS Get a high-speed Internet connection at low costOptions to speed up your connectionThiravudh Khoman
Recently, Database has seen a smattering of news about new high speed Internet services, ranging from cable modems to ADSL to satellite to ISDN to leased lines. Almost invariably these are targeted at corporate users with deep pockets. Given the acknowledged 56Kbps limit of analogue modems, it seems these high tech alternatives are the only way to go faster. But for individual users or smaller businesses with less buying power, there are a few lower-tech/lower-cost alternatives which can help you optimise your Internet access speeds. These aren't the holy grail mind you, but they can help if you're willing to do a little homework.
CHOOSING A WELL-CONNECTED ISP The first step is to choose a "well-connected" ISP. When it comes to choosing an ISP, Internet virgins will usually either buy "blind", buy "pretty", or buy on the advice of friends. The latter isn't such a bad option, but since it's difficult to be absolutely objective I'd recommend adding "science" to your search.
First of all, ask yourself how you plan to use the Internet. If your needs are email only, then practically any ISP will do as long as you stick with POP email or web mail that's hosted at your ISP. Why? Because you're not really connecting to the Internet _ you're simply logging into and using the services of your host's computer only, even if you're sending/receiving overseas email. If your needs include web browsing but due to language problems you plan to use only Thailand-based sites, then you need to see how your ISP is connected to the intra-Thailand gateways. Finally, if you want "the works" and want to be able to go anywhere and everywhere, then you'll also need to see how well your ISP is connected to the world. The way I see it, most people will either fall into the last category or will have their sights firmly set there.
THAILAND CONNECTIVITIES
To determine how well any ISP is connected, first check out Nectec's "Map of Internet Connectivities in Thailand" at http://ntl.nectec.or.th/internet/map/current.html. How do you visit there without an Internet connection in the first place? Use the services of a cyber-cafe or just buy any Internet package that appeals to you and install it. Consider this the cost of a taxi ride on the Internet.
Nectec's map is updated periodically and I recommend that you acquaint yourself with this rather than poring over the advertising that exudes from each ISP's web site. This page contains it all and saves you from having to load graphically-overdosed ISP sites.
On the Nectec map, the thing that you'll be most interested in are the blue "balloons", these being the commercial ISPs. The pink ovals are the intra-Thailand gateways (Public Internet Exchange or "PIE" and National Internet Exchange or "NIX"), while the purple oval is the Communication Authority of Thailand's (CAT) international gateway (International Internet Gateway or "IIG"). The grey bar at the bottom is overseas destinations.
Before we look at a few sample ISPs, there a few things to keep in mind: a) my sample ISPs aren't intended to highlight any especially bad or good ISPs; they're just ISPs that I have used in the past and therefore have some personal experience with b) a bit of computer math: 1024 Kilobits = 1 Megabit; therefore, a 1Mbit line is 8 times faster than a 128Kbit line (1024 = 8 x128).
Let's start with Data Line Thai (DLT, the 4th blue balloon from the left, top row). Assuming that I was logged into DLT and wanted to load a web page hosted at Internet Thailand (ITSC, 1st blue balloon from the right, bottom row), I would have to go through a 128Kbit link to PIE and then a 10Mbit link to ITSC. On the other hand, if I were logged into Asia Infonet (Asianet, 1st blue balloon from the right, top row), I would go through a faster 1.5Mbit link to PIE and then a 10Mbit link to ITSC. Theoretically then, Asianet to ITSC is faster than DLT to ITSC.
Now, let's assume that PIE fails for some reason. To get from DLT to ITSC, you would go via a 64Kbit link to NIX and then a 512Kbit link to ITSC. Asianet meanwhile, isn't connected to NIX at all and there doesn't seem to be a way to get to ITSC if PIE was down. Actually, there is, but it involves going to the US first _ ie, 2Mbit + 512Kbit to the US and then 8Mbit + 8Mbit back up. (Note: Both Asianet and ITSC have two lines each going to North America. Both also use the services of Teleglobe for their overseas links. While Teleglobe is a Canadian company, the links apparently go to the US West Coast).
What about overseas access? To access a US web page from DLT, one would take a 128Kbit link to IIG and then an 8Mbit link to the US. Note that the IIG link is shared by many ISPs and is not dedicated to DLT. Asianet, meanwhile, has two dedicated lines to the US _ 2Mbit and 512Kbit _ while ITSC has two 8Mbit lines. Clearly, both Asianet and ITSC beat the pants off DLT when it comes to visiting the US. And hey, look at KSC _ 34Mbit to the US.
If Nectec's map seems unduly complicated to you, think of the links as roadways, the link speeds as the number of lanes on the roadway, and the gateways as intersections. See it now? TIUG'S ISPWATCH
While useful, Nectec's map is only a static/ideal portrayal. As per the roadway analogy, you won't always get from point A to point B fastest by taking the super-highway; sometimes, it's faster to go by the smaller roads. Why? In a word, "congestion". A slower link with few users can sometimes be faster than a "fat pipe" with an overload of users. Also, don't forget that a link works both ways. A link between Thailand and the US doesn't just include Thai users accessing US sites, but also US users (or even people from other countries) accessing Thai sites.
So, how do we measure "real" conditions? If this were a road, we'd install a traffic counter and compute flow rates. With an Internet link, what we do is to send a data packet from point A to point B and then measure how long it takes to "echo" back. Fortunately, the Thailand Internet Users Group (TIUG) already has a setup to measure this. Visit http://internet.thinet.com/review.html, page down three times and click on the link which says "ISPwatch Results Within the Past 12 Hours". Or go straight to: http://www.inet.co.th/cyberclub/nikornv/ispwatch/.
For the techies out there, what TIUG's test suite does is to start from a fixed point (a major ISP in the US called Digex.Net) and then send a packet to each of the ISPs in Thailand, measuring the round trip times. This is repeated four times, going through each of four main Internet exchanges, namely Mae West, Mae East, Sprint and PAIX.
Now, what does this complicated looking table tell us? First of all, the ISP displayed in the first row is always the one with the fastest round trip time. Interestingly, Chomanan's Worldnet service seems to be a frequent numero uno here, despite not having their own dedicated line to the US. Next, the number in the first column is the average of the four round trip times (last four columns), and of course, the lower the time, the better. Note: If any of the four roundtrips fail, an arbitrary figure of 2000 milliseconds is assigned. This can really skew the average, so keep this in mind. Of course, any ISP which fails one or more of these tests should be viewed with suspicion! The text which follows the ISP name can be a bit confusing, so it's useful to learn the lingo (make sure you have the Nectec map handy when you're deciphering this). For example:
_ Chomanan WorldNet 512KIIG/128KPIE means: a 512Kbit line to IIG and a 128Kbit line to PIE.
_ Loxinfo 2MMCI/5.5MGO/2MNIX/2MPIE means: a 2Mbit MCI line to the US, a total bandwidth of 5.5Mbit via Global One to the US, a 2Mbit line to NIX, and a 2Mbit line to PIE.
_ CS Internet 2MIIG/2MTG/2MSTG/256KNIX/4MPIE means: a 2Mbit line to IIG, a 2Mbit line via Teleglobe to the US, another 2Mbit satellite line via Teleglobe to the US, a 256Kbit line to NIX, and a 4Mbit line to PIE.
The ISPWATCH measurements are run twice a day, and it should be emphasised that these numbers/rankings do _ I repeat _ do change from time to time. For example, several weeks ago, Internet Thailand, with its 16Mbit bandwidth to the US, ranked in the top five. Today (October 15) they're in 17th place. Meanwhile, KSC with its 34Mbit lines is ranked 8th, while last week I saw it in 20th place or so. (Makes one wonder why they aren't number one given their humongous bandwidth?)
In the final analysis, you should avoid relying on a single snapshot of these figures; rather, take daily samples and average them over a period of time, say at least a week.
OTHER ISP CONSIDERATIONS
The ISPWATCH figures, while useful, are again not a be-all and end-all. There are other important factors one should consider when choosing an ISP:
a) Dial-in success rate. For me, this is critically important, especially to my psychological well-being. Perhaps I'm spoiled, but I expect to be able to connect at least nine times out of 10 on the first try. Other people may be more forgiving. Unfortunately, there are no definitive measurements for this type of thing and even guesstimating is difficult, since figures on the number of modem lines and subscribers at each ISP aren't readily available. TIUG discusses the issue as best it can, but leaves the matter up in the air.
b) Quality of your line to your ISP vis-a-vis your telephone exchange. Unfortunately, this is impossible to gauge without actually trying it first. This will often determine how fast a connection you can get, especially with 56k modems. Again, I'm lucky to be able to connect at around 50-52k with a 56k modem, but apparently, not everyone has this good fortune.
c) Cost. Terribly, terribly important unless you've married into a telecom family. TIUG's home page at <http://internet.thinet.com/review.html> has a section ranking ISP packages (page down a bit more). However, it's best that you use this as a starting point only, since it doesn't include all packages available at each ISP.
d) Support. Especially their English-speaking abilities if you're not local born. As a general rule, the bigger the ISP and/or the longer the ISP has been in business, the better their ability to support. In theory, of course. Also, expect most support to occur on the phone or via email.
e) Caching proxy services. A caching proxy server saves web pages which were retrieved by various users of an ISP onto a server/hard disk located at that ISP. If you've set your browser to use this proxy server, your web page request may be returned to you much faster than if you had retrieved it directly from the original site. I'm rather doctrinaire about this and I urge you to eliminate from consideration any ISP which does not have a proxy server. No ifs, ands or buts.
ISP CACHING PROXY SERVICES
Why am I so against proxy-less ISPs? Simply because if you can and do use a proxy server, you can speed up your Internet access with little effort and at no cost. Furthermore, you can conserve bandwidth for other users if you can obtain your pages locally _ just as they can save bandwidth for you if they were to use a proxy server. International bandwidth being an expensive and scarce resource, it should be used optimally and responsibly.
So, how does one go about setting up this beast? Normally, only web browsers (such as Netscape and Internet Explorer) are configured to use proxy servers. If you use Netscape, click "Edit", "Preferences", "Advanced", "Proxies", "Manual Configuration", and then "View". You'll be presented with a list of services. Me, I only set up HTTP (web browsing) and FTP (file transfer) to use proxy services. ITSC is my home ISP and its proxy server, proxy.inet.co.th, is at port 8080.
To setup IE4 to use proxy services, click "View", "Internet Options", "Connection", and then look in the Proxy Server panel for "Access the Internet Using a Proxy Server". Check mark this. For ITSC, the "Address" field is again filled with proxy.inet.co.th and the "Port" field with 8080. - this article continues next page...
|
|
||||||||||||||||||||||||
|
© Copyright The Post Publishing
Public Co., Ltd. 1999 |
||||||||||||||||||||||||||