2001-02-15
6:50:04 PM
-
Company profile -- Applied Metacomputing.
I spoke with Dave Kopans, VP of business development for
Applied Metacomputing.
At a glance:
- Research begun: 1993
- Company formed: 1999
- Funding: first VC round closing now (lead is Polaris Venture Partners)
- Company size: ~19
- Patents: none filed
- Has paying customers? Yes
Applied Meta is not a supplier of CPU cycles, as are the other companies
profiled below. The company is a supplier of infrastructure on which P2P
and distributed-computing companies and products can be built. Applied
Meta makes Grid OS, a next-generation operating system that provides
services for storage, networking, and execution. The company's technology
is the commercialization of the work of Andrew Grimshaw at the University
of Virginia. Grimshaw is the company's CEO.
The company's operating environment, which it calls Legion, is
deployed in a long
list of academic, government, and research organizations.
The company's own research network includes
some 10,000 computers, according to Kopans, and runs calculations in
bioinformatics, finite element analysis, computational fluid
dynamics, and other fields.
Kopans said the company's revenue last year was in 6 figures, which probably
makes it bigger than all the discributed-computing companies combined at
this point.
6:30:58 PM
-
Microsoft exec: "Linux threatens innovation."
Are you sitting down? Have you had your milk and cookie? Do not follow
this link if your blood pressure presents any current danger to your
health. Bloomberg
reports
that Microsoft VP James Allchin has said that freely distributed software
such as Linux could stifle innovation and that legislators need to
understand the threat. Allchin made his comments after Microsoft previewed
it's next version of Windows, XP -- and hours before Microsoft admitted
that it is the target of yet another antitrust probe, this one involving
Corel. Some excerpts:
Open source is an intellectual-property destroyer. I can't imagine
something that could be worse than this for the software business and the
intellectual-property business.
I'm an American, I believe in the American Way. I worry if the government
encourages open source, and I don't think we've done enough education of
policy makers to understand the threat.
I would write more but my keyboard is getting moist from the spluttering.
Thanks to TBTF Irregular and regular columnist Ted Byfield for the pointer.
Updated 2001-02-15, 10:58 pm:
Andrew Leonard commented on this story today in
Salon.
Slashdot also collected lots of
commentary. (I usually browse there
with a threshhold of 3, and for this conversation that shows me 54 of 1020 comments.)
4:20:08 PM
-
Company profile -- Data
Synapse.
I spoke with Jamie Bernardin, co-founder and CTO of
Data Synapse.
He is a physics PhD who went to Wall Street to work as a quant before
starting up the company.
At a glance:
- Founded: January 2000
- Shipped first product: September 2000
- Funding: first VC round closed in late July; no angel money
- Company size: ~30
- Has paying clients? Yes
- What's different: nonintrusive client software
Data Synapse has two offerings, an open-Internet network and a package
sold to enterprises for use within their firewalls. For the
Internet cluster they only accept
computing members who have broadband connections; Bernardin said they have
signed up 70,000 such members. As a deal-sweetener
the company bundles the Zone Labs personal
firewall. They stress the low impact of their client on members' computers:
for example Data Synapse does not store any intermediate results on the
local disk.
The Web site doesn't make it clear, but I believe Data Synapse does not
now do computations for paying customers using the open-Internet cluster.
Their pro-bono work involves the research efforts of NCSA. As payment for
the use of their computers, members get
Flooz funny-money. Members can
donate their Flooz to a variety of charities through
Charitableway.
The Data Synapse enterprise offering, dubbed WebProc, makes use of idle
computers within a company's own network. Data Synapse specilaizes in
distributed-computing solutions to problems involving risk management,
such as derivatives calculations for the financial-services industry.
The company has one paying customer, two in negotiation, and 6 in beta.
2:22:03 PM
-
Bill Joy's next J.
Sun's chief techie Bill Joy gave this morning's keynote at the O'Reilly
Peer-to-Peer Conference and introduced -- in the vaguest possible terms --
what he's been working on and what's next. Following up on Java and Jini,
the project is called Jxta
(pronounced juxta) -- Joy recited the now-familiar difficulties in
naming, trademarking, and reserving a domain name for any new project --
and Sun would like your help on it.
Jxta was described as an open-source effort to develop a minimal infrastructure
layer that various P2P clients and projects can use. Joy listed a few areas
in which he expects Jxta to define primitive functions to enable P2P interactions:
- network pipes between peers
- ad-hoc grouping of peers
- monitoring / metering P2P interactions
- security, identity, anonymity
Moderator Tim O'Reilly
asked the blunt question of Sun's proprietary intentions for Jxta. Joy replied
(I thought convincingly) that he hoped Jxta would provide needed services for
some of the components of Sun ONE (their .NET competitor), but that the project
would in truth be entirely open. Joy has chosen the Apache model for both open-source
development and BSD-style license.
Updated 2001-02-16, 10:54 am:
Jxta seems to have very little presence outside Sun at this point.
Jxta.sun.com does not exist. Both jxta.org and jxta.net are
registered to Sun. The former leads to a login screen and the latter is, at this
writing, unavailable. Jxta.com has been reserved by a woman with a San Francisco address
and no visible connection to Sun Microsystems; it gets DNS from servers
in Hackensack, NJ. She may be in line for some largesse soon. (Thanks to Bill St.
Clair for some of this research.)
Bill Joy said that the Jxta project plans an open Web-based event in April
(no more exact date was given) at which more details and plans will be
unveiled. The April event will also serve as a public call for participation
in the open-source project. Of the attendees at Oreilly's conference Joy
asked that they mail jxta@sun.com if interested in getting involved
early.
2001-02-14
8:34:13 PM
-
KnowNow lifts the veil.
Rohit Khare, CEO of the secretive KnowNow.com,
talked for the first time in public this afternoon about what the company's
technology will enable. The room was SRO.
KnowNow is developing what Khare calls the
Two-Way
Web. Others use that term (Dave Winer for example). Here's what KnowNow means by
it:
By creatively reusing existing HTTP standards with client/server pairs at
each node, developers enjoy real-time notification, a standard namespace,
and most importantly, a zero-install, scriptable development platform:
the 4.x browser.
Among the demos that Khare flipped through was a voting application that
updates in real-time, anywhere on the Web, with total votes cast. The
application required less than 600 bytes on disk. (That is not a typo.)
The audience wanted to see more of what the KnowNow technology can do --
the company is rumored to have a suite of 30 demo apps guaranteed to drop
the jaw of anyone exposed to them -- but Khare was done.
KnowNow's product offering will be introduced at PC Forum late next month.
Updated 2001-02-15, 11:10 am:
Dan Gilmor, who was also at Khare's talk, has
spilled the beans
to a greater extent than I was willing to do yesterday. Here's how he explains
the key to KnowNow's juju:
It turns out that a Web browser can hold open the connection to the server.
Normally, a browser sends a request for information, which is delivered by
the server. The connection ends. KnowNow holds the connection open. Then it
adds some JavaScript and, voila, you have a mini-server inside the browser.
You're not necessarily using lots of bandwidth, but you are pretending, in
effect, that you're downloading a very, very long document while the browser
keeps communicating with the server.
Updated 2001-02-15, 2:20 pm:
Rodger K. Crawford sent this URL
from Google's cache that points to some of the information KnowNow used to have up on their Web site, before
they decided to pull the covers up tighter. KnowNow was
Slashdotted
last Friday when word got out that Wilfredo Sanchez had left Apple's Darwin project
to join the company; in that thread some enterprising Slashdotter revealed the Google
cache trick.
7:14:38 PM
-
Company profile: Parabon.
I spoke with Jim Gannon, CTO of
Parabon Computation.
At a glance:
- Founded: June 1999
- Public beta: June 2000
- Shipped first product: November 2000
- Funding: 2 angel rounds, VC round in progress
- Company size: ~50
- Has paying clients? Yes
- Has patents? One filed: bizmeth patent for the process of
selling the idle time of distributed computers
- Client: Java + native code
- Server: Java
- What's different: emphasis on security; product maturity
Parabon sells two products: Frontier, an ASP-like service utilizing Net computers
for computation, and Frontier Enterprise, targeted at corporate use inside a
firewall.
Like Popular Power, Parabon emphasizes the two-way security of the Java
sandbox on client desktops and strong protection for the intellectual property
of its paying clients. Parabon code running on the client is obfuscated such
that it is difficult (thought probably not impossible) to figure out what the code
is doing. (The company uses a commercial Java obfuscator.) Parabon also guarantees
its customers anonymity: the computation provider does not know what company's
code s/he is running. (Some potential providers object to this feature, because of
moral or ethical concerns; Gannon said that the number who do is small enough
not to impact their marketing efforts.)
Parabon's pro-bono computation, which runs on providers' machines
when no paid work is available, benefits
cancer research.
The company signed its first paying customer a week ago, in the financial
services industry. Parabon has submitted a purchase order to this customer, so
strictly speaking to date the company hasn't received any revenues. A number
of customers are in trials for the ASP service, Frontier, but so far none
has signed up for cash.
Parabon currently runs a sweepstakes for the benefit of computation
providers. Gannon said that soon they will begin a program of direct
payments. A typical PC might earn $5 or $10 a month.
4:39:06 PM
-
Company profile: Popular
Power.
I spoke with Nelson Minar, CTO of
Popular Power, one of a number
monetarized distributed
computing companies with a presence here at the P2P conference.
At a glance:
- Founded: January 2000
- Shipped first product: April 2000
- Funding: Angel round (investors include Tim O'Reilly, Eric Schmidt, and Brian Behlendorf)
- Company size: ~20
- Has paying clients? Yes
- Has patents? Filed
- Client: Java + native code
- Server: Java
- Distinction: emphasis on security
Popular Power targets a number of vertical industries such as pharmaceutical,
software testing, VLSI testing, and financial. Their distributed application
server provides a simple script-based framework for running off-the-shelf
problems as well as an API for detailed customization. The company sells
servers mostly into corporations who want to put their idle computers to work.
Servers are priced based on the number of cients supported.
Minar expressed surprise that his is the only talk at the P2P conference focusing
on security in the P2P realm. Popular Power distinguishes its offerings by the
degree of trust they offer to both paying customers and clients. By using the
Java sandbox, Popular Power can guarantee that computation customers won't
misbehave and steal data from client computers. And the company has devoted
significant resources to the (hard) problem of protecting the customers' data
and computations from the spying eyes of those operating its clients.
Popular Power runs a non-profit project on
optimizing
flu vaccines. The company does not now use the clients over the open
Internet for paid computation jobs, but is committed to developing this
ability. Minar said that solving the problems posed by the untrusted
environment teaches lessons that can be applied within corporate intranets.
4:30:45 PM
-
Counting computers and
predicting the future.
Clay Shirkey, one of the organizers of
this conference, ended his talk on the lessons of Napster this morning with
the following killer closer. Maybe you've heard it, but it was new to me.
T. J. Watson, fifty years ago, allowed as how he could forsee the need
for maybe five computers in the world.
We now know he was wrong. [Pause for audience chuckles.]
We now know he overestimated by four.
Shirkey continued, "I don't know exactly what it will look like,
but it won't be Sun ONE and it won't be Dot NET, though both of
those will be part of it. Whatever it looks like, the people
who will make it happen are almost certainly in this room."
The O'Reilly P2P conference is sold out at about 850 attendees, of which
150 are journalists. (When Tim O'Reilly asked audience members to identify
themselves, far fewer than 150 confessed to this ink-stained, or
electron-stained, calling.)
10:46:52 AM
|