|
|
|
|
|
|
Vyatta Interview
We believe that open-source has created a profound change in the market place, shifting the balance of power from large corporations back to the users of products. Open-source is about cost savings, increased flexibility and choice, and better security. At
Vyatta, we're trying to bring those qualities to the world of
network
infrastructure products.
They are the
kind of product just like the Windows firewall that runs on somebody's
desktop.
While products like this can be valuable, they only protect a single machine. While the can be used on servers, they
can be difficult
to
administer when you need to make a rule changes that affects many
machines. This argues for the standard firewall appliance, essentially what you call a "hardware firewall." A firewall appliance sits in the network and applies security policies to all traffic flowing through the device. You can firewall a whole data center full of servers with a single device. This makes it easy to update when policies need to change.
Rather than changing every server, you
simply
update the rule set in the appliance. Guarddog
is a firewall configuration GUI. It isn't a firewall
itself, but allows you to configure other firewalls more simply. Basically,
you can run Firestarter on another distribution of Linux and
it will
do the firewalling. In some ways, Vyatta is similar to a firewall appliance or Firestarter when used in a gateway mode: we are a gateway device that firewalls all the traffic flowing through us. Where we're really different is that we focus on the whole of the gateway problem. Put another way, the market for routers and firewalls is merging. People finally looked around and noticed that just about every company has a firewall and a router at the connection to the Internet. The market is responding to consolidate those devices.
If you
look at
what Cisco is doing with the ISR series and Juniper/Netscreen is doing
with the
ISG series, it's clear that these functions are converging. So,
Vyatta is really following that trend, supporting both
strong routing and firewalling features. Whereas a Netscreen product
might have
good firewalling, but be weak on the routing, or Firestarter doesn't do
the
routing at all, Vyatta is a complete product with both functionalities. We're both a consumer as well as a contributor to open source projects. The Vyatta system is based on a customized distribution of Debian GNU/Linux. We make use of the Linux kernel's fundamental networking capabilities and then layer a set of protocols such as BGP, OSPF, and DHCP on top of that. We tie it all together with a CLI and a GUI that a network manager would understand. To do all that, we rely on some components that are developed outside as well as some that are developed inside. Moving forward, we have been growing our development community.
Because the product is based on Debian, there are a wealth of
opportunities for people to get involved and integrate functionality
into the system.
hackers mailing lists, a wiki, an open bug tracker (Bugzilla), and
a community Knowledge Base. These are free resources.
These both provide updates and tech support through the subscription period (1-year, 2-year, and 3-year options).
The difference between
Professional and
the response time guarantee to any issue.
Vyatta's web
site has more detail.
We have a couple of automated test beds here at Vyatta to which we're constantly adding tests. We have dedicated test engineers who exercise the product in ways that the automated testing just can't and are also responsible for adding to the growing test cases. For things like hardware compatibility testing, we perform some of it ourselves for selected platforms, but then we rely on the community to increase the breadth. There
is so much diversity of hardware out there
that we
can't hope to replicate it all ourselves.
with two different goals. First, there is a large community that generally wants the latest features. Second, there is a body of paid users that want a slower rate of change but greater stability.
To address
these groups,
In this model, the community version becomes a fast-paced development version where new features are introduced regularly. The commercial version is the object of more focussed testing. Features
propagate from the community version to the commercial version as they
are
stabilized. Bug fixes propagate from the commercial version back to the
community version as problems are identified. Essentially,
the main differences between the two are: * The community version will generally have more features. We're just implementing this strategy right now, however, and we're a bit out of sync on this. Currently, the subscription edition has VPN support,
which is not in the community
version, but this should change shortly. * The subscription edition will generally receive more testing and will be more stable at any point in time for a given feature. This is not to suggest that the community version will be unstable,
but only that it will be the first place
where
features appear and then mature. *
Finally, with
the
subscription edition, you're guaranteed that you can get support from a
trained
expert as opposed to taking your chances on the mailing lists.
To really address the market of home users and make money doing it, we need a *VERY* low-cost hardware platform. That means porting to something other than x86.
We
can do
this, but we'll need some hardware partners to help with a clean,
low-cost,
easy-to-manufacture hardware design.
They are really
addressing the home users, running on the existing WRT hardware
platforms.
We're going after the small-to-medium business market where things like
hardcore routing are more desired. Again,
the reason
is that it requires a lot of changes to our build system and we'd
probably have
to cut down our system dramatically to make it fit in the limited
resources
provided by those hardware platforms. This is a deep, structural change. While companies like Cisco make good products, they charge an astronomical price for them. For instance, we had one user look at expanding his existing Cisco router with a single Fast Ethernet port to do some additional LAN segmentation. Cisco charges $1400 for a single Fast Ethernet port for the 1800/2800/3800-series routers! In my world, that's a standard PCI card at $20. This
customer
realized
that he could buy two Vyatta subscriptions for the price of that one
Fast
Ethernet port. At
Vyatta, we have taken the stand that a standard network
administrator, familiar with products from Cisco, Juniper, and others,
should
be able to sit down at a Vyatta system and basically feel at home.
Rather than
dropping the user at a standard Linux bash shell prompt and asking him
to edit
configuration files with vi or emacs, we provide an integrated CLI like
you
would find on a "normal" networking system. While we do allow a user
to drop down to the bash shell to do something advanced, the normal
operation
of the system doesn't require it. This means that users don't have to
become
Unix system admins in addition to network admins. A lot of the
knowledge they
already have transfers over to Vyatta immediately.
|
News
|
||||||||||||||||||||||||||||||||||||||||||||||||||
|
Objectis is a Pilot Systems and Blue Dynamics service
This service is supported by Zope Village and Plone.org CPSSkins is Copyright © 2003-2004 by Jean-Marc Orliaguet. Plone and its visual design is Copyright © 2000-2004 by Alexander Limi, Alan Runyan, Vidar Andersen. |