<<<As I look across this littered babbling technology landscape I see an unending amount of confusion as to how best to resolve these issues of connection and bandwidth.>>>
The drums for Gigabit Ethernet are starting to beat a little more loudly each month. Imagine, the throughput of a mainframe on the network, then SERIOUS client/server starts. The below is part of the article from InternetWeek dated 01/05 (not as MSFT biased as ZD publications), we couldn't have dreamed of this unbiased report even 6 months ago, the worm is turning. (By the way, we just installed the latest Service Pack on our you-know-what, now the e-mail is spontaneously crashing throughout the network, and my IT friend is gone)
****************************************************************
The increase in the number of Gigabit Ethernet switches will encourage a growing market for network interface cards. Existing Gigabit NICs, such as those from Packet Engines and Alteon, will be joined by others. For example, Intel is preparing to release a new single-chip Gigabit Ethernet solution in 1998. The increasing competition, and improved yield rates from transceiver manufacturers, should reduce NIC prices during the year.
ÿ
A major missing factor in Gigabit Ethernet is reserved bandwidth. Currently the purview of ATM, Gigabit Ethernet vendors such as Extreme Networks are working on a standard to support this capability. You can expect Extreme (whose switches will remain purple), and others, to deliver comprehensive reserved bandwidth solutions this year in a form that should interoperate with other vendors' products.
ÿ
Of course, ATM already provides reserved bandwidth as well as a true multimedia networking capability. Coupled with the growing support for OC-12's 622-Mbps bandwidth, you can expect ATM to grow in acceptance during the coming year. In fact, the primary limitation on bandwidth won't be the network infrastructure, but rather the network operating systems and the server hardware that supports them.
ÿ
For example, only Micron currently supports a 66-MHz (twice the throughput of current 33-MHz PCI bus, jww), 64-bit PCI bus on its high-end workstations and servers. Likewise, network operating systems don't all measure up to the speeds required for this year's new networks. While Novell NetWare and most flavors of Unix can handle gigabit speeds with aplomb, Microsoft Windows NT can't handle the load. Network managers needing actual throughput will need to look beyond Microsoft for the foreseeable future.
ÿ
Operating Systems
ÿ
1998 isn't a year to look for massive improvements in operating systems. In fact, if you're hoping for solutions to the shortcomings of Windows NT, you might want to take this year off. Although Microsoft has been promising NT version 5.0 for delivery sometime in '98, there is growing doubt among many observers that this will actually happen.
ÿ
Adding to the woes of managers of NT-based networks, version 5.0 won't provide any perfor-mance enhancements, so Microsoft's sleepy-time speed limits will remain in place. The expected lack of improvements will be exacerbated by Microsoft's own difficulties with the U.S. government, which are likely to so consume the company's attention.
ÿ
Though the outlook is gloomy for NT users, things aren't really a lot better elsewhere. Unix users can expect to see incremental improvements and relief from the Microsoft hegemony since, for the moment, Unix will be a safer buy.
ÿ
Unfortunately, not every version of Unix, nor every Unix hardware provider, will be likely to see 1998 as a great year. Though Sun is seeing its market grow, both in its bread-and-butter SPARC platform line and its Intel-based product line, others, such as Silicon Graphics, are not expected to show significant changes. On the other hand, smaller companies, including Caldera, continue to bring out new solutions to make life easier for those that need Unix, and the bandwidth Unix can support, including such things as built-in support for SMP servers.
ÿ
Caldera, incidentally, also continues to support and improve DR-DOS, and is releasing 32-bit Web browser support (called Web Spider) for this platform, which is also being positioned as an alternative means of Internet access for thin clients. Meanwhile, Caldera is working to link NetWare, Unix and other operating systems on the corporate network. The company will have a version of Netscape's SuiteSpot for the open Linux market, and it will be introducing NetWare for Unix during 1998.
ÿ
The biggest opportunity in 1998 lies with Novell and its new version of NetWare, code-named Moab. When this product debuts as NetWare 5.0 in mid-1998, it will support TCP/IP as its native network protocol. This, plus continued growth of NDS, places Novell in an enviable position, provided Big Red can turn its position to advantage.
ÿ
Meanwhile, Novell is talking about its native support for an implementation of a Java Virtual Machine that's aimed specifically at allowing NetWare to be a Java hosting product. And, of course, the company is making much about its native IP support. However, Novell's biggest opportunity will be as a high-bandwidth, easy-to-manage alternative to Windows NT.
ÿ
Novell is going the extra mile in its battle against NT with its new, easy-to-use graphical installation tool. More important to managers will be NetWare 5.0's Java-based GUI management utilities, it's ultra-high performance BorderManager (the best proxy cache available, jww), and broader support for Domain Name Server and Dynamic Host Configuration Protocol within Netware Directory Services.
ÿ
Otherwise, the greatest changes in operating systems will be in new management innovations, and in better support for bandwidth demands. You can expect to see native support for Fibre Channel-based storage, gigabit networking, Ultra-2 SCSI and new high-bandwidth processor and bus technologies. These improvements will couple with the emerging high-speed networking infrastructures to provide pathways for the vast new demands on bandwidth.
ÿ
In short, managers can expect to see a wealth of new tools that will let them merge their corporate networks with the Web, to make the Internet an integral part of their intranets, and to have the bandwidth and infrastructure required to make that happen. For companies that are able to look beyond today's hot topic and see that tomorrow brings new and different opportunities, 1998 is nearly without limits.
*********************************************************************
jww |
