Gus, You get the SI, Son of SAN foreshadowing award for '02. DN told us how BRCD really really wanted MCDT before it got spun out. Perhaps some of the reasons why are now on the table... from the archives of S of S.
Gus, thanks for the time you put in on the thread.
From last summer....
Biting the hand(s) that feed you has a short shelf life as a business strategy.
Well put, Douglas.
Additionally, it's becoming clear to me that Brocade's product roadmap is reflecting the unusual weakness of its patent portfolio, which currently consists of a grand total of 1 utility patent and 2 ornamental enclosure designs generated in more than 6 years of operation.
The significance of a patent portfolio, of course, lies not only in its royalty income and cross-licensing potential, but more importantly, in the way it serves as a foundation for new patents that result in new products.
For instance, Brocade's only utility patent contains references to 9 McData patents, 2 IBM patents and 1 Hitachi patent. It is fair to infer that Brocade's 5 other patent applications will have to contend with the prior art established by McData's comprehensive patent portfolio which currently consists of 22 utility patents (including two ESCON/FICON related patents) and 16 patent applications.
There's a very good chance that in order to overcome the prior art of McData's patents, Brocade will have to modify its applications with major design workarounds that only add unecessary design complexity that ultimately affect performance.
McData's crossbar switch patent, for example, is quite comprehensive and practically precludes Brocade from coming up with a competitive non-blocking director switch. This is probably why it is trying to position its unproven core switch in an imaginary class above directors. Typical Brocade antic.
There are two other areas of weaknesses in Brocade's current product lineup worth mentioning.
First, the easily verifiable fact is that the 2400, 2800 and the 6400 DO NOT support non-disruptive code downloads. Brocade is claiming that its 12000 will support non-disruptive code downloads, but that only means that a SAN based on Brocade switches will suffer from the unnecessary complexity of having switches that support and do not support this feature, a schizoid network design if ever there was one. In contrast, all of McData's directors and fabric switches already support this feature.
The significance of non-disruptive code downloads, of course, is that it allows a SAN vendor to flex the codes of the HBAs, switches and storage devices on a SAN to adjust to the multiple changes constantly being made in the server platforms, operating systems, middleware, applications and clustering services. As the CTO of EMC was recently quoted as saying: Interoperability frequently requires re-engineering.
Secondly, Brocade indicated that it plans to have the 12000 ready for alpha and beta testing by the spring and summer with volume production by October. This was one of the most amusing part of the earnings call next to Reyes's mealy-mouthed contention that customers actually wanted to replace high availability non-blocking director switches with less reliable blocking switches and vaporware that is not even in the alpha-testing stage yet!
I think what prompted this particular line of rhetoric was the joint EMC-McData announcement that they can already use the E-Port standard to provide interoperability between Brocade and McData switches. In practical terms, this means that while Brocade is trying to manage a major and complex technology transition, there is nothing to stop either EMC or McData from moving in on a Brocade SAN and replacing the less reliable Brocade switches at the core with high availability directors while redeploying the displaced Brocade switches at the department and workgroup levels.
I also think the frantic way Brocade is trying to introduce too many new technologies at the same time is the clearest indication yet that it is feeling the heat from its OEMs and the competition. In haste there is often waste.
Consider that not only is Brocade trying to introduce a new ASIC technology at 1 Gbps and 2 Gbps -- by itself a major endeavor fraught with risks -- it is also trying to introduce an unproven switch platform loaded to the gills with first generation software features. It is well understood that in highly interdependent parallel systems, the software features are only as good as your hardware. Put another way, it is impossible for an unstable switch platform to produce useful SAN services software.
Again, for contrast, McData is already shipping 5th generation directors and will probably have its 6th generation directors up and running by the time Brocade starts shipping its first generation core switch, a switch which for all the hoopla around it will not even match the director switches in terms of non-blocking features, security and reliability.
What goes around comes around. |
