Shootouts at PCQ Labs are normally exhaustive, and more often
thannot, leave the team doing the shootout pretty exhausted as well. But
frankly, this beats our previous records hollow.
The searing heat of the Indian summer was enough exhaustion
in itself. Forget the scheduling nightmares–coordinating couriers, air
tickets, and railway timetables–for eight people and three PCs traveling
across nine cities. To that, add the round-the-clock tests that went on well
past midnight, checking out transfer rates, and harassing unsuspecting helpdesks
with preplanned questions.
But it sure had its asides. Like when one helpful helpdesk
guy wanted to know whether Krishna Kumar had any technical knowledge. Meanwhile,
Suma was having fun trying to make sense of the torrent of Telugu coming from
the other end at Hyderabad, while Vini was trying to cope with gems like
"your 56 kbps modem will not work on a 33.6 kbps line" at Chennai. At
about the same time, Anil was at his wits’ end trying to buy a 25-hour MTNL
account in Mumbai. Now, let’s proceed to how we tested the services provided
by the ISPs.
Which cities?
This wasn’t too difficult. The four metros selected
themselves. Pune made it to our list as the first Indian city to embrace IT.
Bangalore and Hyderabad too entered the list on the same count. Now, we wanted
two other cities–one from the north, and another from the south. From the
south, the choice was Trivandrum–the capital of Kerala, the only other
southern state, and from the north, it was Chandigarh–the capital of both
Haryana and Punjab.
Which services?
We set out to test all ISPs who offer dial-up services in the
selected cities. However, we had to drop a couple of services during the
testing, because we couldn’t acquire an account during the testing period, or
because they offered only year-round accounts and had no smaller accounts.
Getting the account
All accounts used for testing were purchased off the market,
or as in the case of a couple of ISPs in Hyderabad, were acquired for
"checking out in order to buy" without disclosing our actual identity.
We bought the lowest-priced account the ISP had to offer.
The test bed
The tests were done using similarly configured PCs running
Win 98, and connected to Ace 56 kbps modems. The telephone lines used were
direct lines–that is, they weren’t EPABX extensions. In each city, the
operating system was freshly installed on the machine.
The tests
We went the FTP route to check out data transfer rates.
Though we did check out browsing speeds, they weren’t used in the final
reckoning. This was because if FTP transfer rates were good, HTTP-based browsing
speeds were invariably good; but this wasn’t true the other way around. We
repeated the tests during various times of the day, ranging from early morning
to past midnight. We dialed into each ISP for a maximum of five attempts during
each round. Once connected, the same file–which was on a server outside India–was
FTP’d five times. The average of the transfer speeds, as well as any wide
variations, have been reported. With many ISPs in the process of setting up
their own gateways, more comprehensive tests were deferred for later.
Calls were made to the helpdesk of each ISP, both during
regular working hours and during the night and early mornings. Questions ranged
from where to buy the account to whether roaming was possible to whether it was
legal to use Internet telephony software.
Pricing
One thing we haven’t considered in this shootout is the
cost of an account. There are two reasons for this. First, prices are rapidly
falling and have changed many times even as we were in the midst of the tests.
Secondly, the cost of an Internet account has fallen to a significantly low
level that you can now start looking at the level of service–ease of
connecting, transfer rates, etc–rather than at price.
Anil Chopra, Ashish Sharma, Krishna Kumar, Lalit Juneja, Sachin Makhija,
Sandeep Saxena, Suma EP, and Vini Goel