
| February 23, 1998 | ||||||||||
IP Multicast: Is It Ready For Prime Time?An interview with Judy Estrin, CEO of Precept Software
By
Jason Levitt
InformationWeek:
The 1998 IP multicast Summit showcased some
interesting case studies of IP multicast deployment from
organizations such as Federal Express, NASA, and UUNet.
Despite their success with this relatively new technology,
it seems most companies are still testing the waters. What
has been the major factor(s) that has kept companies from
going for it? What technologies or issues are changing the
climate for IP multicast?
Estrin:
What initially slowed widespread acceptance was a
lack of understanding of just what IP multicast could do, a
lack of motivation because applications didn't exist, and a
lack of stable implementations by infrastructure vendors.
Over the last year the landscape has changed dramatically.
In the application area, we now have streaming video and
multicast file-transfer products (such as those from Precept
and Starburst). In the infrastructure arena we have
Microsoft's TCP protocol stack and Cisco's routers with
ma
ture implementations of IP multicast.
But, over the same period, the Internet and corporate
intranets have emerged as mission-critical tools -- and,
with good reason, corporations are cautious about
introducing any new technology into their mission-critical
environments. So IP multicast is being phased in rather than
implemented all at once. The IP Multicast Initiative, with
nearly 100 member organizations, is helping to drive this
process. And only recently have we seen a spate of both
users (NIH, Federal Express, the University of Oregon) and
content providers (Disney) publicly demonstrating the
viability of multicasting technology.
InformationWeek:
IP multicast has practical applications for
corporate sites with large numbers of users, but is it
practical, from an ROI perspective, for smaller companies to
deploy? I'm afraid your answer might be, it depends on the
application. If that's true, can you give an example where
you think a smaller company would reap significant ROI from
IP multicast?
Estrin:
Multicast is applicable in any organization with a
need to communicate information from one to many (where many
can be 10, 100 or 10,000). Though in larger organizations
the investment is higher, greater bandwidth savings per user
will be realized. In smaller environments, on the other
hand, multicast can be easier and less costly to deploy
because there is likely less equipment to upgrade. In a
fully switched network, for example, there are no routers to
be upgraded; in a flat LAN with no switches OR routers, the
issue is even simpler.
The applications for IP multicast will vary based on the
size of the organization. Whereas a 50-person firm can
conduct leader-led training in a room on site, a company
with thousands of employees spread out across the globe
would find IP multicast an ideal vehicle for delivering
"distance learning." On the other hand, using IP multicast
to bring television (e.g., CNN and other news programming)
to employee desktops would
be just as effective for the
small company as for the large one.
InformationWeek:
Despite the bandwidth-saving attributes of
IP multicast, widespread deployment over the Internet seems
like it could cause major congestion, especially if
broadcasting of audio and video becomes easier because
Internet service providers have enabled IP multicast. Do you
agree? What form do you see IP multicast taking over the
Internet?
Estrin:
There's no question that transmitting video and
audio over the Internet does create a bandwidth demand that
wouldn't otherwise be there. But now that users have seen
how effective such transmissions can be, the demand for them
will only increase, whether or not IP multicast is deployed.
IP multicast is the only technology that can address this
situation effectively by reducing the number of video/audio
streams actually being sent. By the year 2001, IP multicast
will be commonplace on the Internet. In the meantime, it
will be phased in gradually, first
with tunneling,
satellite, and other interim technologies, and ultimately on
its own.
InformationWeek:
IP multicast deployment brings up a lot of
security issues for the enterprise. Routers have to be a lot
smarter to handle IP multicasts, new software may have to be
deployed on desktops, and access to IP multicast sites
outside the firewall are a possibility, as well. What do you
feel are the primary security risks for corporate sites
deploying IP multicast? Are any standards being developed to
handle them?
Estrin:
Any new technology brings with it new concerns of
security and management. IP multicast per se has no unique
security issues. However, the applications that use IP
multicast (e.g., streaming video) typically generate UDP
(Universal Datagram Protocol) rather than TCP traffic.
Historically, most firewall products have not had mechanisms
for selectively dealing with UDP traffic. While no standard
ways to do this have emerged, firewall vendors, because of
the
sharp increase in UDP applications, have recently begun
to develop products that can cope with various multicast
applications. Corporations will have to be alert to make
sure that the firewalls they purchase are capable of
protecting all the mission-critical applications they have
deployed.
InformationWeek:
IP multicast is widely touted as the
solution for broadcasting bandwidth-intensive data such as
video, audio, and large files, even though it is not yet
widely deployed. Do you see any new classes of applications
appearing once IP multicast is widely deployed? What major
consumer applications do you envision?
Estrin:
It is as yet unclear whether any wholly new classes
of application will emerge to take advantage of IP multicast
once it becomes ubiquitous. However, three application
categories will become much easier and more efficient to
deploy (and therefore far more widespread) once they can
escape the heavy bandwidth burden of unicast. One is
audio/video, includin
g streaming programs, video-on-demand,
and multipoint conferencing. A second is information
dispersal, including multipoint file transfer and
information updating (e.g., stock updates and other "push"
applications). And the third -- the one we've heard least
about because it's the least sexy -- is a set of system-
level functions such as simultaneous updating of Web caches
and replication of databases for redundancy. Obviously a
major consumer application for IP multicast will be
entertainment, as content providers send everything from
feature films to multiplayer interactive games over the Net;
Disney recently became the first major player to announce
development of a multicast channel. But electronic commerce
will likely play just as large a role in multicast, with
videos that let a prospective home buyer watch a video of
new houses on the market, or a prospective car buyer see a
demonstration of a new model's road performance.
For more information about IP multicast, visit the IP
multi
cast Initiative Web site at
www.ipmulticast.com
.
|
||||||||||
Videology Imaging seeking Software Architect in Greenville, RI
Beyond.com seeking Database Developers in King of Prussia, PA
Mentor Graphics seeking Sr. Director of Sales in San Jose, CA
Mesalands Community College seeking Comp Sci Instructor in Tucumcari, NM
Sectoral Asset Management seeking IT Manager in Montreal, QC
For more great jobs, career-related news, features and services, please visit our Career Center.