Messages in this thread | | | From | (Scott Lurndal) | Subject | Re: [NEWS STORY] Crap on MSNBC | Date | Tue, 6 Apr 1999 11:10:58 -0700 (PDT) |
| |
> Has anything read the trash at > http://www.msnbc.com/news/256197.asp > > Choice lines like "But he said Linux currently lacks some of the features > demanded by corporations that intend to run their entire business on > computers. Among them are the ability to run simultaneously on many > processors in a single computer and to keep a log of what the computer > has done." > > G'day! <I keep thinking SMP and syslog. I don't know what THEY are > thinking though.>
Before you too quickly jump on MSNBC, you should be aware that there are places where linux needs work before it would be suitable for enterprise IT shops. While linux (2.2.3) does support multiple processors, it doesn't scale very well past two or three processors with workloads that include a significant amount of I/O activity. Several factors, including the serialization of SCSI subsystem queuing across all processors, acquisition of the kernel lock in 'sys_read' and 'sys_write' all lead to a non-competitive scaling factor per processor. Note that all of these problems are expected to be addressed in some way in the future.
From a log standpoint, many commercial IT shops require auditing facilities much more extensive than that provided for by SYSLOG and it's ilk. For example, some bank auditors require complete audit trails for a system including an audit record for each file open, file removal, file creation, process creation, privilege level change, process deletion, etc. Not many commercial unices have this ability (except perhaps some that provide a C2 audit trail may have enough information). Most mainframe operating systems provide this level of detail in an audit log. As an example, the Unisys V-Series mainframes provide three logs:
SPO log: (log of interaction between operator and system) Run log: (log of process create, file create, file delete, etc) Maint Log: (log of single event upsets, disk sector retries, etc.)
The SYSLOG on linux corresponds most closely to the SPO & Maint log above, while an audit trail on C2 systems would correspond most closely to the Run log. The maintenance log was especially useful to the Field Engineers responsible for preventive maintenance as it allows the determination of trends (i.e. if a particular device was retrying frequently, the heads could be cleaned, media replaced, etc).
At some point, I believe that the linux community will be required to decide the target point for the linux operating system; will it be the desktop? or will it be the small departmental server (file/print serving), or will it be the large SMP server? In many ways, the goals and design philosophies of each target point differ; and attempting to cover all targets in a single product may become somewhat unwieldy. (which brings us back to the fork, not-to-fork discussions at the last Linux World Expo).
scott
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |