Now that Ancor's stock has appreciated, I've been motivated to do more
due diligence. Consequently, I have been placing a lot of phone calls
to catch up with former colleagues and associates to find out the full
skinny with FC these days. I have included some facts/observations by
folks really in the know. Furthermore, they have no motivation to hype
a stock. For a variety of reasons, these guys are not allowed to
own stock in Brocade, Ancor, EMC, or Vixel.
Each point which I list below represent a distillation of hours of
conversation and have been confirmed by two or more experts in the
small world of SAN engineering. Of course any technical mistakes are
mine since I could have misunderstood the lingo (pretty deep at
times). I have attempted to keep the jargon and technical content to
a bare minimum in keeping with the SI forum. Please correct if I have
taken too great of liberties with the truth.
I apologize in advance for the random order. I just
transcribed my scribbles written on napkins :-)
--------------------------------------------------------------
* Brocade and Ancor are definitely the top contenders regarding
switches. Everyone seemed to agree that the competition is very good
for the industry. One company pushes the other to innovate and take
technical risks.
* There is a philosophical difference between the approach each company
pursues regarding switching technology. Speaking simply, Ancor is
primarily hardware-based while Brocade relies more on software
(firmware, actually). The split has some ramifications:
A. Ancor will usually set the upper-bound in switching
speed since pure hardware should always be faster.
---I was cautioned, however, that given the whole
scheme of things, the theoretical edge may evaporate
with real data sets. If you move very large files,
the speed differential should be minor. Conversely,
move lots of smaller files and Ancor's advantage
could weigh heavily.
---Approach could make for cheaper manufacturing once
ramp up gets to appropriate levels.
B. Brocade has a lot of flexibility since firmware
is pretty easy to change. Furthermore, the Brocade
switch was designed to be configurable in such
a way that other switch vendors could use parts
for specific designs of their own. In other words,
the chip count issue can cut both ways---not necessarily
in Ancor's favor.
---I found two camps on the wisdom of this approach
as well. Brocade gets other competitors' juices flowing
because their arrangement lowers barriers to entry for
other designs. That is to say, Brocade may be
giving away the store by providing "building blocks"
instead of going a pure hardware route.
* I could not find a single person on God's green earth who claims to
have a legitimate application for Class 1 (other than Ancor's Terry
Anderson who I did not speak with). Class 2 seems to be required for
tape drives due to reliability.
* Both companies have changed considerably over the last two years
from a management point of view. Brocade has the slickest management
which was so aptly exhibited by their IPO timing and admirable
handling of the announced Sun loss. Manufacturing ramp-up represents
an entirely different challenge, however.
* Both Brocade and Ancor switches work to a high level of performance.
I found a guy in the movie industry using Ancor switches who swears by
the product. Conversely, Brocade clearly has a reliable and sturdy
switch as well. Some OEM vendors found Brocade's engineering staff
easier to work with primarily because it is much bigger than Ancor's.
We sometimes forget the obvious.
* Insinuations that Ancor won Sun's business by seducing them with
warrants are sour grapes. Brocade would have killed for that
contract. When I tried on Roy's "Most Favored Nation" argument
regarding "equivalent" OEM's like SGI and IBM, one expert started
laughing so hard he became momentarily incoherent. Once I settled him
down, I made a run with the "San Jose vs. Minneapolis" home turf
advantage. Quick work was made of that one when he wrote on the
blackboard the locations of Dell, EMC, Data General, Gateway, etc.,
etc.
* Very few labs have both Ancor and Brocade hardware configured to
enable head-to-head comparisons. Those that do are reluctant to
broadcast for a number of reasons. So much integration support
depends on good relations with respective engineering staffs,
there is no productive purpose to serve.
* Even the top-notch facilities have little practical experience with
cascading FC switches yet. Again, the competition between Brocade and
Ancor can be generalized as falling into two camps. The high
performance computing folks favor Ancor because they seem oriented
towards larger fabrics. Brocade has been working the lower port count
side of the street such as found with NT workstation vendors.
Companies like SGI and IBM are schizophrenic since they support both
NT and Unix workstations. Sun is going the Internet and larger SAN
configurations.
* Another reason why you do not see bake-offs between Ancor and
Brocade switches is that the two not work together in cascades as far
as I can tell. However, just a few days ago, there seems to have been
some significant agreement that the switch behavior will be
interchangable soon. This is probably good news for Ancor!!!!
--- Continue with another message ---- |
