What happened to Ashok Kumar? He was an early supporter of Ancor when Ancor was misguidedly positioning Fibre Channel as the ethernet-killer. Now he is just a shrill advocate for one questionable IP storage start-up after another while never missing an opportunity to declare that fibre channel is dead. In the process of grandstanding, he reveals his ignorance of the distinction between the architecture (SAN) and the interconnect (Fibre Channel).
Understanding that difference would have helped him understand the historical progression from first generation SANs (based on the half-duplex ESCON interconnect) to second generation SANs (based on the full-duplex Fibre Channel interconnect) that is now underway.
Being aware of that historical progression would have helped him gain some insights about the actual learning curve of customers who separate storage traffic from general purpose network traffic by deploying a deterministic network (SAN) behind a probabilistic network (LAN). That chronicled learning curve is one of the keys to understanding how the different vendors are segmenting their target markets.
Instead we get grandiose statements like the ones below from a PC analyst who still has a hard time accepting the inexorable shift from conventional wisdom (processor-centric) to the unconventional wisdom (storage-centric-->information-centric):
February 15, 2001
....Storage is a necessity and has a strong future in the long term, but it is also an industry that has been overvalued. Fibre Channel has shrunk too far to be a major force again.
news.cnet.com
April 5, 2001
.....iSCSI products are starting to come to market, and while they won't compete immediately with Fibre Channel SAN switches, it is only a matter of time before they do. After iSCSI will come InfiniBand, and then there will be additional pressure on Fibre Channel as DAFS technology is delivered. Where is the excitement and push to revive Fibre Channel? Those days of excitement and mystery are over. Even if these other technologies take a couple years to launch, what is a year or two in the big picture of infrastructure technology developments? Fibre Channel is already as good as dead--as sure as FDDI is...
news.cnet.com
August 15, 2001
...Our channel checks indicate that SAN implementation continues to remain anemic due to weak IT spending, glut in storage capacity, and lack of compelling TCO benefits....
IT spending is anemic? No shit, Sherlock!
However, to try to use that to support an argument when the facts are not there is just plain sad.
Below are the growth rates of the key switch vendors. The numbers clearly show that McData, Inrange and Brocade continue to grow; albeit, at a much slower pace. In no way do the numbers show the kind of anemic growth that one would expect if these companies were selling a product with no compelling TCO. By the way, CMNT just reported an 100+% sequential increase in SAN over WAN revenues. That's about as anemic as they come.
Message 16216490
Regarding compelling TCO benefits, here are three reports from EMC, Brocade and McData. Self-serving? Yes, but no less so than the reports of an analyst promoting one damn pre-revenue company to another while only half-understanding what he is talking about.
How Three EMC Customers Lowered Total Cost of Ownership
emc.com (IDC)
Reaping the Benefits of SANs: Impressive ROI and Improved Operations (KPMG)
brocade.com
Migrating to SANs: A Report From The Front Line (IDC)
mcdata.com
This is a closer look at how one customer is actually deploying its SAN:
.....Michael Obiedzinski, vice president of Information Systems at DTCC, said that since starting the project 10 months ago, he's continued a cautious adoption of the SAN model, beginning with the company's tape backup systems and more recently integrating its disk arrays. Mission-critical data related to daily clearing and settlement activities will remain on its legacy mainframe system.
"We put servers on the SAN that we considered to have an internal impact first," he said. "Some applications will be on our mainframes for years to come."
DTCC is moving to the SAN in response to enormous growth in data related to increased trading. For example, over the past four months, data has increased fourfold. Obiedzinski also hopes to reduce the 30% to 40% of available data space that is wasted by inefficiencies in the current architecture.
An advantage of the SAN, which is about half completed, is an increase in the data throughput related to backup speeds, which have increased from 30MB per minute on 4mm tape drives to 400MB per minute on the Fibre Channel network. DTCC is also now able to manage four times as much data with the same number of data administrators.
The project includes consolidating 350 to 400 servers throughout the company into two data centers for ease of control, a move that dramatically reduces backup times.
The DTCC is using industry-leading technology such as Fibre Channel switches from San Jose-based Brocade Communications Systems Inc. and directors from McData Corp. in Broomfield, Colo. The DTCC's storage boxes include Hopkinton, Mass.-based EMC Corp's Clariion FC4700 disk array, Louisville, Colo.-based Storage Technology Corp.'s L700 tape silos and Sun Solaris and Intel Corp. servers running Windows NT and 2000.
Once the top-of-the-line equipment is in place, Obiedzinski wants to add more automation to his systems.
"The next thing we will move toward will be snapshots," Obiedzinski said. "We'll take snapshots of the data and backup from that so we won't have to tie up application servers."
Obiedzinski said the SAN is attached to redundant data centers, one in New York, the other in an unspecified off-site location. The data centers are connected by dark fiber using Dense Wavelength Division Multiplexing technology, which increases bandwidth by shooting different data streams at different frequencies or wavelengths over the same optical fiber wire.
computerworld.com |
