Article and comment: (FreeBSD) Daemon's Advocate
Hu, it seems some FreeBSD people really got up the tree after
some recent statements about performance of NetBSD, and in the
January issue of DaemonNews, there's an
entry in the "Daemon's Advocate
column from Scott Long and Robert Watson from FreeBSD where they
try to address the "NetBSD > FreeBSD" claims recently assessed
in various benchmarks, and some of the statements from Christos Zoulas
in the annual report. I'd like to present my personal views on a few of
their statements here:
So to sum up my views on this article, it seems to me that
the first part looks like a rather bad researched rant
with the hope to spread enough FUD to make people not look at NetBSD,
which I'd like to welcome everyone:
features list (needs updating),
- ``As the old saying goes, FreeBSD is about performance, NetBSD is about platform portability, and OpenBSD is about security.
So is that still the case?''
I don't think so, and it probably never was, as I hope for FreeBSD
(and OpenBSD too): Each of the BSD projects has more than one
goal, and at least for NetBSD, establishing state of the art
security is just a natural thing that needs no special emphasis.
Likewise, if performance deficiencies are found, the past has shown
that they do get addressed. As such, asserting that each of the
BSDs has only one goal in mind seems a short-sighted to me, and
leads to wrong impressions in the reader that e.g. performance
and security are of no concern in NetBSD. Which is wrong. Of
course NetBSD is performant, and of course it is secure - we just
emphasize something else as our major feature.
Speaking from a historic perspective, when UCB stopped BSD,
FreeBSD and NetBSD emerged (with some detour via 386BSD).
With 4.4BSD supporting a wide range of hardware platforms,
FreeBSD concentrated on the PC platform only, and NetBSD
set the goal to provide a stable Unix(like) platform that
behaves equally well on all platforms (as far as the platform
allows). Seems NetBSD didn't have to re-adjust its goals
- ``The NetBSD advocates are quick to claim that NetBSD 2.0 now beats FreeBSD in both performance and features. Fortunately, that is just not true.''
There are a number of benchmarks out there that show differences
to this statement. Not all of them, not in all details, but the
statement in general is not true any more.
- ``There simply is not anything else in any other OS that is as flexible, easy to use, and full-featured as netgraph.''
Uhu. While reading the description on what Netgraph is, I wonder:
What about System V's STREAMS, which was there um... 20(?)
- ``Advanced network features and protocols such as SACK, NFSv4, SYN-cache/SYN-cookies, compressed TIME_WAIT, and accept filters allow for fast, secure, and scalable network operations in an ever-increasing hostile and busy Internet. Packet filters like IPFW and PF provide advanced filtering, shaping, and NAT sharing.''
Um, just for the record, NetBSD has SACK too, SYN-cache etc. is
available since NetBSD 1.3(?), a collection of packet filters (IPF,
PF) and traffic shaping. I'm not sure what the other features listed
mean, so I cannot comment. (Someone please update the NetBSD list
of features please!)
- ``Outstanding desktop and laptop support is provided by a number of technologies.''
Um, like the PCMCIA framework that I hear is based on NetBSD's code?
Given it's goal of portability, NetBSD was long made to support a lot
of devices, even in a "removable" way, and thus getting
drivers working esp. on non-Intel, 64-bit machines helped a lot.
I agree that interfacing with some vendors like nVidia or groups
like KDE and GNOME could be improved, but then KDE runs pretty well
out of the box on my PC running NetBSD. :)
- ``The "Ports" collection provides one-step support for over 11,000 3rd party applications.''
Yes, size does matter. :) I hear rumours that many of these
"ports" are duplicates for difference in options and language,
and that some do not build properly. To counter this with some
number, from the ~5500 packages in pkgsrc, less then 100 (<2%) were
broken in the last attempt to build them all.
- ``NetBSD 2.0 is a significant step forward for NetBSD, but the large amount of stagnation cannot be overlooked.''
- ``It's great that NetBSD is committed to supporting legacy architectures, but how does the effort to do so benefit modern architectures or encourage wider use and more adoption of NetBSD?''
Portability is more than "supporting legacy architectures", as
be known. It's about making hardware abstractions, and
establishing interfaces to easily replace support for code, like
change of CPU or bus architecture. This lead to products like
the ARM-based "keyboard-video-mouse" switch from Avocent,
the PowerPC-based Brocade SAN switches of the Sony PSP TCP/IP
stack, just to name a few. And it also serves as a base for
FreeBSD's PCMCIA/Cardbus support, FreeBSD/alpha, FreeBSD's
bus_space/bus_dma(?) support, I hear. :)
various NetBSD-based products from HP, IBM and Sony,
products based on NetBSD)
Maybe some more understanding of the issue and facts at hands
should be gained before making more statements like the above.
- There's one statement in Robert Watson's (second) part of the article
that I want to emphasize though (in favour of FreeBSD):
``This includes doing a better job with PR.''
I have promoted NetBSD at various joint BSD-boots at roadshows in
Europe, and while NetBSD and OpenBSD were present, there were always
people asking for FreeBSD which we had to send away. It would be
really nice if some FreeBSD people could be found to do some on-site
adovcacy at various events. Anyone interested feel free to contact me
to coordinate joint BSD activities!
Last, I'm impressed that DaemonNews gave their good name to publish
such a statement in such a position.
[Tags: Articles, rants]