Re: FW: The term "operating system"

[Randy Thornton <thor@sprynet.com>]

  PMFJI,
  
  I can summarize some details from the article for you:
  
  1) Oranges & old Oranges
  
  They compared NT 4.0 SP3 versus NetWare 4.1. Well, NetWare 4.1
  is TWO versions old (Jan 95) & the latest version - IntraNetWare - 
  came out around the same time as NT 4.0 shipped (about a year ago).
  
  2) Test Server 
  
  (page 64) "server ... a generic clone ... 120 Mh Pentium, 64 Meg RAM,
  a NE 2000 nic, a master 2.1 gb hard drive and a slave 2.1 gb EIDE hard
  drive, with the slave drive partitioned in two equal areas - 
  one for NTFS, and the other for Novell's file system."
  
  3) Test clients
  
  Six clients (hardware unspecified) running NT workstation 4.0. 
  For the NetWare test, they use the Microsoft Client Service for NetWare
  with Microsoft's SPXLink (ie Microsoft implementatin of IPX/SPX).
  
  4. Test software
  
  The testing was done with something call Dynameasure Enterprise 1.5 from
  Bluecurve (http://www.bluecurve.com), which simulates the stress of
  multiple users on a single machine. Each client si
  
  5. The test
  
  They tested only filesharing performance, essentially copying files to
  and fro, measuring average response time, "motors" per step and total
  throughput. A "motor" is a single instance of a simulated user running
  on the client, which is running mutiple motors at the same time to
  simulate multiple users. (The test seems highly dependent on the
  testing software.)
  
  
  
  Such a scenario is of course very unlike any real production network. 
  
  More interestingly, the Bluecurve testing software is designed to specifically
  simulate capacity tests on NT. It is really a scenario deployment tool 
  for NT and SQL server. If you visit the Bluecurve web 
  site you will find out they are MS Solution Provider Partner and 
  doing a tour with MS:
  
  "BluecurveÂ’s Dynameasure product line represents a whole new class of software
   tools, and a new way of thinking, of how to deliver a systemic approach to
   performance management on Windows NT," said Chuck Dietrick, General
   Manager, Microsoft Northern California District. "By partnering with
  Bluecurve
   on this educational seminar series, we are providing the marketplace with a
   turnkey solution to many of the questions they might have on deploying and
   managing their Windows NT infrastructure."
  
  
  There is nothing wrong with this and this is not to say they product is not 
  good: capacity planning for networks is a very very useful tool for 
  implementation planning, especially for NT, but it obviously was 
  not designed to do cross OS benchmark tests.
  
  
  At 05:30 PM 11/14/97 -0500, chuck wrote:
  >
  >Christopher Pall[SMTP:x97pall@wmich.edu] writes:
  >
  >>>In a similar vein, the latest WindowsNT Magazine has an article
  detailing their lab tests
  >>>pitting NT head-to-head w/ NetWare 4.11. The only reason I'd take their
  conclusions >over
  >>>yours is that they detail exactely how the experiment was done, what
  equipment and
  >>>software was used, what was measured and how the data was interpreted.
  One of the
  >>>principles of science is that I could recreate and independantly verify
  the experiment
  >>>and it's results.
  >>
  >>To think, a windows NT magazine found NT beating Novell software, wow,
  must have been a tremendous
  >>leap for them. Where is the URL for this comparison? I'll help you read
  through their "results"
  >
  >Heh, a very legit response! But like I said, anybody (with the time and
  resources) can do the
  >tests themselves.
  >
  >http://www.ntmag.com/issues/1997/Nov/Nov97toc.html
  >
  >  - but they haven't posted the data on the web yet. The bastards want us
  to go out and buy the atoms.
  >
  >
  >
  --
  Randolph W. Thornton
  President, The LAN Guide Company
  www.languide.com