TO: ICANN
NEW TLDs divisionBy e-mail
By certified mail
November, 13-th,
2000
RE: Abacus America, Inc.
Application for new TLD Registry operations
for
.biz; .cool; .fam; .inc; .xxx
Response to ICANN staff report http://www.icann.com/tlds/report/biz1.html
A. General
We
believe that a part of our application was NOT properly classified. The statement
that “… its application focused primarily on the .biz TLD “ is simply not true.
Throughout the application, we paid equal attention to all names. It was only
because we listed these names alphabetically that .biz became the first name on the
list. While we definitely support .biz, we believe that the four (4) other names
proposed by Abacus are equally strong. Our belief is supported by the independent
poll conducted by MSNBC, where four (4) out of five (5) of our names are listed in
the top 10 names voted upon by thousands of participants.
We request hereby that
our application be reviewed again. Moreover, we ask that our proposed TLD .fam
be included in the “personal” group where it belongs and that .xxx is included in
the “restricted content” group where it belongs.
B. Technical
We are stunned
by the fact that Abacus – the only applicant among 44 who offers a live working prototype
of the full TLD Registry system http://rodopi.abac.net is so strongly undervalued
from a technology standpoint. We believe that this is a misunderstanding and we will
gladly provide further information about how our systems will scale and perform under
load. Furthermore, we are applying for five (5) domains which generate 32 possible
combinations of approved TLDs and 96 total possible combinations at 10% 50%
and 90% level of confidence. Obviously there is no way for one system to fit all
these possible combinations, which vary in size. We described the basics in our proposals,
whereby additional power could always be added to our basic solution. The selected
operational systems UNIX for DNS and MS 2000 Data Center for back office operations
are both scalable and cluster able. Furthermore, we conducted significant testing
in our own labs and our engineers came to the conclusion that the systems proposed
by us will match and outperform our business proposal and will scale upwards well
beyond that. We will gladly open our labs to ICANN’s advisors to re-confirm our conclusions.
One of the reasons our systems perform so well is the hardware implementation of
SSL protocols.
We take the technology very seriously and that is why we prepared
an itemized response to all technical issues.
a. Total Capacity. “The equipment
appears less well configured than other proposals for similar TLDs”
Response:
Unlike
the configurations proposed by other TLD applicants, our configuration was tested
in the prototype environment and under simulated load. Our tests showed that the
DNS servers could handle a load much higher than the average load on the current
root DNS servers for .com, .net and .org TLDs. The described servers configuration
can handle a load higher than the load projected in the business plan. Nonetheless,
we are able to upgrade the system if and when needed since scalability and cluster
ability are built-in our proposed OS. Also, our configuration includes hardware SSL
implementation, which will result in much better performance compared to RRP systems
with software SSL implementations running on similar servers. We will gladly share
our testing model and results with ICANN’s technical experts.
b. Projected Growth
Rate. “The machine architecture described in the technical plan does not appear
to be designed to scale up rapidly. The architecture described relies on individual
dual processor systems running Microsoft Windows 2000. Preferable choices are
clustered systems or hosting on a platform that scales to larger configurations through
simple substitution of other computers.”
Response:
Latest performance tests
show that Microsoft SQL Server 2000 outperforms all other database solutions, including
Oracle (see http://tpc.org). Microsoft Windows 2000 Datacenter scales up to 32 CPUs,
64 GB memory, and a 4-node clustering and load balancing solution and this offers
more than enough scalability for the database servers. The Rodopi R architecture
allows for an unlimited number of RRP servers working simultaneously with the database,
so additional RRP servers can be added at any time. The Whois server can also scale
to a 4-node clustering solution.
c. Startup Period. “The technical plan projects
10,000 registrations on the first day, 100,000 in the first week, 1 million in the
first month and 3 million in the first quarter. Were the applicant to be granted
all of the TLDs being requested, the land rush period is likely to see far higher
volumes than these. The systems do not appear to be sufficient to handle the
startup period.”
Response:
The numbers described in the business plan are not
the technical limitations of the proposed systems. We estimate that the servers described
in our proposal can handle at least 1 million registrations for the first 2 hours
(150 registrations per second). Again, this estimate is supported by extensive lab
tests under simulated load. We will gladly share our test model and results with
ICANN’s technical experts.
d. Fault Tolerance. “The fault tolerance strategy for
all machines except root DNS servers is to keep spare machines hot. The fail
over strategy was not described. There was no provision for a second geographic
location for servers other than DNS servers.”
Response:
We will replicate the
main registry location at all locations originally described as DNS locations. Each
location will be able to take over registry operations at any time. All changes in
the registry database will be mirrored by the Rodopi servers at all locations. This
mirroring technology is currently in use by the “Rodopi Roaming Club” to synchronize
roaming accounts and usage data between more than 600 Internet Service Providers.
3. Apparent
Implementation Risks. “The proposed Registry software, Rodopi R, has not been deployed
in a production environment. The infrastructure appears to be undersized for the
land rush period, and the architecture does not appear to allow for easy scaling
up.”
Response:
At present, the only registry with software deployed in registry
production environment is Network Solutions. Rodopi R is based on the commercial
product Rodopi, which is used by more than 600 Internet Service Providers and Telcos.
The Rodopi DN and Pirin software products are based on Rodopi as well, and are currently
used commercially by one Registrar and hundreds of Resellers. Abacus America, Inc.
is the only new TLD Registry who offered a working prototype (http://rodopi.abac.net
), based on a commercial grade product.
C. Business review.
We were similarly
surprised by the Staff Report’s comments about Abacus not having the marketing expertise
or understanding of the markets.
First of all Abacus generates over $10 million
in annual revenues. Such a figure could not be achieved without significant understanding
of our markets and a great deal of marketing expertise. Our marketing budget is close
to $2 million. We have prepared several business plans lately for variety of reasons
and we know that short business plans are more valued than long ones. It was
for this reason alone that we avoided stating obvious truths and “reinventing the
wheel” about Marketing in our TLD application.
Second, Abacus was the ONLY applicant
having to have been able to predict four (4) proposed TLDs which ended up within
the top 10 list in the independent poll by MSNBC. That prediction was made
after extensive market research (see Abacus’ TLD Application, whereby we stated,
in part, that the reason for selecting each TLD proposed was because of our own market
research.
As of Oct, 12, 2000, the tally by MSNBC was as follows (current results
at: http://www.msnbc.com/modules/surveys/domainname.asp ):
1. .xxx (1783 votes)
proposed by Abacus
2. .web (1668 votes)
3. .sex (1281 votes)
4. .biz
(1115 votes) proposed by Abacus
5. .inc (997 votes)
proposed by Abacus
6. .kids (947 votes)
7. .news
(825 votes)
8. .fam (594 votes) proposed by Abacus
9. .shop
(536 votes)
10. .info (471 votes)
No other applicant came close to predicting
the demand for registrations in any way that approached Abacus’ market research.
The closest applicant had only two (2) domains in the top 10 list. Results such as
this serve as independent verification of Abacus’ knowledge of the markets, the segments
we are serving and the general marketing principles.
Third. Abacus’ entire business
model is 100% focused on the Internet. It sleeps, eats and breaths Internet. What
better way of knowing Internet trends, markets and segments than the real day to
day hands on experience that comes with this kind of access and exposure to the daily
needs of Internet users?
Fourth. Abacus has experienced strong growth of over 100%
per year for the last 3 years. Such growth cannot happen without clear business and
marketing vision.
Respectfully, Abacus therefore disagrees with the characterization
of its marketing and business abilities. Nonetheless, Abacus is committed to
working with ICANN in every way to satisfy ICANN that it will, in fact, provide lasting
security and stability if it is chosen as a Registry.
D. Comparison with other
applications.
.biz
iDomains – only 4 employees; does not have history and
experience; controversial issues including some ICANN activists, who are personally
involved and CORE, involved with this application as a subcontractor as well as a
principal in other applications. Abacus’ application is definitely much stronger
in regard to the technological and business experience and capabilities.
JVTeam
– It seems they are not a team anymore. It is official now that NeuStar and
Melbourn IT split due to conflict of interest, because Melbourn is involved in competing
application. With the team falling apart, NeuStar does not have relevant technological
experience and technological back up.
With all due respect to the financial power
of Martin-Marietta and the general experience of NeuStar, Abacus has much more relevant
experience and working technology, which makes it a better candidate.
KDD – Financially
strong applicant and adds geographical diversity too. The most serious of its problems
is that it is partnering with NSI, thus reinforcing NSI’s monopoly in TLD markets.
Abacus’ application is better because it does further add to the problems of this
monopoly. Many postings in the public forums show how sensitive the people are about
monopolies. ICANN was created to break the monopolies, not to reinforce them.
.fam
Abacus
is the only applicant for this domain name listed #8 in the top 10 list by MSNBC.
.cool
Abacus
is the only applicant for this popular name among the young generation. The
comments in the first ICANN public forum bear this out.
.xxx .inc
We reserve
the right to analyze the competition on those domains when the names are approved
by ICANN.
E. Conclusion
As shown above, most of the comments evaluating Abacus’
application can be addressed by further submissions to ICANN. If there are
other portions of the application which did not reflect these points, Abacus is eager
to augment any of those sections to the satisfaction of ICANN. We will cooperate
with ICANN by providing further evidence if and when needed.
Abacus has spent
the last two years in preparation for successfully managing new TLDs. We
created strong technology and have built our business model around the Internet and
providing service to the Internet community. Abacus wants to do everything
in its power to demonstrate that it is the best choice for the TLDs which it has
proposed.
We are looking forward to working with ICANN to establish reliable and
stable Registry operations.
Sincerely,
Ivan Vachovsky,
CEO
Abacus America,
Inc.