OS News

24 August 1998

Be OS and the Enterprise

By David Choi

It seems things are going quite well for Be.  
In describing the Be OS technology, buzzwords abound  
It does not take a networking genius to see that the Be OS has many of the underlying OS technologies for great networking  
There is another area where the Be OS' technology may help it excel: clustering software.  
Presently the company does not have the resources to focus in on yet another aspect of OS capability  

Be, the OS and the company, have come a long way since the beginning. But they have certainly been through their fair share of trials and tribulations. For example, after Apple chose to acquire NeXT's technology over Be's, it seemed they had hit some dire straights. However, soon thereafter, a $25 million dollar injection of venture capital funds had Be humming along once again. Be continued to aggressively pursue its vision for an operating system, first for the PowerPC and then for the x86 architecture. They released timely updates and improvements, while whetting the public's appetite with demos that showed off the OS's technology and ferocious speed. And, in version 3.0, Be had a fairly successful launch of their first retail version of the software.

It seems things are going quite well for Be. They have a flourishing geek community rabidly following the company's every move. And, sometime in September, the next retail release of the software, version 4.0, should be out. So then, why do I feel like there's something missing?

From the company's infancy Be's target market was made very clear. This target, digital content creation, has been steadfastly adhered to by Be. And with this goal in mind Be has done a marvelous job of creating a technologically sophisticated OS, without the legacy baggage of other OSes. In describing the Be OS technology, buzzwords abound: protected memory, true preemptive multi-tasking, pervasive multi-threading, 64 bit journaling file system, symmetric multi-processing, etc. These characteristics are crucial for an operating system geared towards content creation. Today, special effects film makers can have 3D files that require several days to render, even on high powered Alpha systems! Therefore speed and stability are obviously very important. However, these technologies are not just requirements of the visual effects industry. Networking, especially at the Enterprise level, has also required these technologies for quite some time

A large multinational corporation's network can have literally tens of thousands of computers located in various cities and countries. Keeping a network of this size working smoothly is a monumental task. The requirements of an Enterprise level OS are in some ways even more rigorous than those of a media OS. An Enterprise level OS must be compatible with a myriad of hardware and network protocols, not just high end graphics devices. It must be robust, read: not prone to crashing, because in these environments a network failure of just a few hours could mean millions of dollars lost in revenue. Also, because servers can be handling thousands of requests at one time, a network operating system must be fast and scalable.

It does not take a networking genius to see that the Be OS has many of the underlying OS technologies for great networking. However, it does have a long way to go before it can be considered ready for the Enterprise. For example, hardware compatibility and network protocol capabilities are sorely lacking. Currently Be OS has only TCP/IP and Apple Talk protocols implemented. Also utilities and other tools for monitoring and administrating a network are simply non-existant. So obviously the Be OS is not going to become an enterprise class networking package any time soon. However this does not detract from the potential of the OS for the Enterprise, since architecturally so many high end features already exist.

There is another area where the Be OS' technology may help it excel: clustering software. For those of you who are not familiar with the technology, clustering software essentially allows the servers on a network to act as one giant supercomputer. It is analogous to multi-processing, but clustering uses entire systems instead of just additional cpu's.

The average supercomputer can cost upwards of a million dollars. Even large corporations find it difficult to make that kind of investment on a computer. What clustering does is allow an institution to leverage its existing computer hardware infrastructure to do highly complex data processing. Imagine for a moment a stock brokerage that is doing a historical commodities analysis that dates back several hundred years. Obviously this type of analysis could be extremely processor power hungry. Now imagine this same company, instead of purchasing a multi-million dollar supercomputer, using its existing computers and network to do this analysis. The needed work could be completed without the high expense and trouble of new hardware. On top of creating a network supercomputer, clustering software can enhance system availability. Since by definition a cluster contains multiple servers, if one server should go down another would be immediately available to take its place.

So what is it? Why doesn't Be take advantage of the OS's networking potential? Well there may be several reasons for this. Presently the company does not have the resources to focus in on yet another aspect of OS capability. It probably has its hands full developing the media side of the OS and bringing general hardware compatibility in line. Also the head of Be, Jean Louis Gasee, is an ex-Apple executive. Apple was and still is one of the premiere content creation OSes. Therefore the skills of Gasee's leadership are honed in that area.

When Be ported its OS to the Intel architecture it was both booed and applauded. However, whether you philosophically agree or disagree with Be's decision to do the port, from Be's perspective it was only being realistic. After all, roughly 95% of all desktop computers run on x86 processors.

Be's decision to make its OS hardware agnostic can also be applied to the functions of the OS itself. Let's be realistic - the content creation market is a moderately sized niche at best. In order for the company to make the real money (yes I said the m word) it must open its OS up to broader markets. From an architectural or technology perspective there is nothing to hinder the Be OS from making in-roads into the home/gaming and networking environments. Heck, with its compact size, Be's OS would be ideal for use in palm tops and embedded devices - even game consoles!

The only hope, at least in the immediate future, for high-end networking to work its way into the Be OS is through third party software developers. Fortunately, thanks to Be's great developer support, companies willing to make the commitment to the Be OS will receive a wealth of help and guidance. Also, the OS' modular architecture should make it relatively easy to add new networking features.

All in all, Be's potential for success exists in spades and it seems only a matter of time before some entrepreneurial company takes advantage of this, or Be itself makes another first move.

David Choi can be reached at davchoi@earthlink.net

Home | News | Features

Copyright © 1997 OS News