Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : The New QLogic (ANCR) -- Ignore unavailable to you. Want to Upgrade?

To: Greg Hull who wrote (25976)	2/3/2000 12:45:00 PM
From: Lhn5	Read Replies (2) \| Respond to of 29386

Craig, reread George's post about bias and research. He has spent his entire adult life reading and evaluating research papers, and probably writing a few as well. I have done the same. There are many conflicting results from various very credible sources studying many issues. It is the readers job to sift through it all with a suspicious mind and come come to conclusions if possible. You cannot read a research report and say..."oh". George is right. You are wrong.

To: Greg Hull who wrote (25976)	2/3/2000 12:56:00 PM
From: Smart_Asset	Respond to of 29386

Some light reading on coming attractions. is.pennwellnet.com Article Date: January, 2000 Magazine Volume: 4 Issue: 1 -------------------------------------------------------------------------------- CONNECTIVITY: Prepare for the shift to switched fabrics The road from PCI's shared-bus architecture to InfiniBand switched fabric architecture is through PCI-X. By Thomas Heil Switched fabrics are a revolutionary advance in server architecture. While new to certain markets, this architecture has existed in mainstream enterprise computing for many years. The concepts of switched fabrics are embodied in the current InfiniBand (formerly System I/O) initiative, which reflects the best of the Future I/O (FIO) and Next Generation I/O (NGIO) standards. The proposed standard promises to make systems and I/O attachments simpler, cheaper, more reliable, and easier to expand and service. However, the transition from shared-bus architectures to switched fabrics will take time. PCI is the dominant shared-bus standard and is well established. For switched fabrics to emerge, the industry must allow a graceful transition, implying a potentially long period during which buses and fabrics will coexist. PCI-X-an enhancement to the PCI bus that is expected to emerge next year-ensures the ongoing viability of the PCI bus architecture and provides a non-disruptive path from bus-based systems to fabric-based systems. What is switched fabric I/O? In contrast to shared buses such as PCI, switched fabrics connect processor and I/O nodes via an "any-to-any" network of switches and point-to-point links (see figure). Nodes communicate using intelligent messaging and protected remote DMA (RDMA) techniques, in contrast to bus-based operations in and out of a shared memory. Click here to enlarge image Switched fabric I/O standards Switched fabrics are not new. Proprietary fabrics have existed in mainframe, massively parallel, and Unix midrange markets for years. What is new is the intent to define industry standards that enable the fabric to eventually replace the bus as the I/O interface in high-volume markets such as Intel architecture systems. Today, Fibre Channel (switched fabric and arbitrated loop topologies) is emerging as the standard interconnect for storage area networks (SANs). Ultimately, however, a system area network standard-one that can handle all I/O functions-is needed. Switched fabrics offer the following promises: A "unified" system area network, in which a common switched fabric handles all I/O functions, including cluster inter-processor communication (IPC), network, and storage. A limited distance focus emphasizing "intra-chassis" or "same room" connectivity. (Fibre Channel and LAN/WAN technologies will handle long distance applications.) Incorporation into host chipsets to eventually eliminate PCI slots. Definition of a new "fabric I/O adapter" that plugs into a switched backplane instead of a shared bus. Fabric I/O adapters will be hot-pluggable and externally accessible. Bus-to-fabric transition The transition from bus-only architectures to fabrics will probably be gradual (see figure on next page). The Bus-Only curve represents the transition starting point-i.e., servers with PCI slots only. A Bus-Only server will connect to a fabric with a PCI-SIO host adapter. The Bus-Fabric Hybrids curve represents servers with PCI slots and native (integrated) fabric channels. The Fabric-Only curve represents the transition end-point-i.e., servers with no PCI slots where all I/O occurs over native fabric channels. Bus-Only systems will dominate the early years, while switched fabrics (implemented via PCI adapters) are introduced, debugged, and proven in production environments in all the major operating systems. During this time, a peripheral base and support infrastructure will emerge. Although some early-adopter OEMs may provide native fabric channels sooner, the broader market is typically reluctant to integrate technology that is not thoroughly proven or is not widely available. PCI slots are a low-risk proving ground. Once fabric technology is proven and a reasonable peripheral base exists, OEMs can increasingly ship Bus-Fabric Hybrids. Over a period of several years, OEMs will shift the mix from "mostly PCI" to "mostly fabric" as the fabric peripheral base grows (similar to the way OEMs used PCI-ISA hybrids to move the ISA adapter base to PCI). This process will continue until PCI connectivity is no longer required. Click here to enlarge image PCI-X provides a bridge PCI-X is a significant step in the evolution of PCI. It increases bandwidth with higher frequencies and protocol enhancements as well as scalability with electrical improvements that enable OEMs to provide more high-frequency slots. PCI-X and other initiatives like hot-plug and alternative form factors will significantly extend the useful life of the bus architecture while maintaining legacy compatibility. Switched fabrics are compelling, but it's clear that bus architectures will play a critical role well into the next century, and PCI-X provides an evolutionary path. As switched fabrics emerge, the role of Fibre Channel and SCSI may change somewhat. Today's PCI-SCSI and PCI-FC adapters will gradually give way to fabric adapters. These adapters will enable server vendors to replace PCI slots with native fabric channels, while providing connectivity to Fibre Channel SANs and SCSI or Fibre Channel peripherals. Switched fabric I/O is the architecture of the future, with potential to deliver significant value to end users. It is, however, a fundamental architectural shift that will unfold over many years. Just as important to end users is a non-disruptive path to this new architecture, which implies a period of bus/fabric coexistence. Thomas Heil is a senior systems architect at LSI Logic (www.lsilogic.com), in Milpitas, CA. PRINT SCREEN

To: Greg Hull who wrote (25976)	2/3/2000 1:12:00 PM
From: Joe Wagner	Respond to of 29386

Interesting post by Bill Fischofer on the EMC thread. About evolving data centers and the high margin products becoming the center of the network. I think he is on the right track, and this bodes well for the FC Directors & switches that will populate the data centers. Message 12765093

To: Greg Hull who wrote (25976)	2/3/2000 1:23:00 PM
From: Craig Stevenson	Respond to of 29386

Greg, If I had 1000 shares of BRCD, I would sell it. The valuations of Brocade seem astronomical to me, even given what I see as their gorilla status. If I were a BRCD shareholder and had seen the KeyLabs test, I would want to see the unpublished test results that DIDN'T favor the SilkWorm, and I would want to know what management intended to do about it. I would also want to know why my CEO was apparently lying about certain things. And if he said he wasn't lying, I would want to see the cold, hard facts that supported his statements. I would make sure that management would see other companies, especially Gadoox and Ancor as fierce competitors, and I would pound into Brocade employees that they can't be complacent and overconfident. I think Gadzoox threatens the low end, while Ancor could threaten the low and high ends of the market. Of the two companies, I would see Ancor as the greater threat. Sun provides instant credibility as soon as shipments begin in volume. I would in no uncertain terms, want to make sure that Brocade had a director class switch ready to go in very short order. (If the SAN market heats up more quickly than predicted, demand for director class products could materialize, and Brocade would not have a product for that market.) "(I suspect you've never owned an equity about which you had no concerns or reservations <g>)" Correct. <g> I tried to make the point in an earlier message that I'm not just singling out Ancor here. I do this to EVERY company that I'm involved in. (Even though I'm not in ANCR, I still follow the company tangenitally.) Although you did not ask, if I had 1000 shares of ANCR, I would hold it. The upside potential seems greater for Ancor than for Brocade, since the disparity is so great at this point. Craig