Eskimo North


          [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

          Upgrades


          • To: eskimo-announce@eskimo.com, outages-list@eskimo.com
          • Subject: Upgrades
          • From: Robert Dinse <nanook@eskimo.com>
          • Date: Fri, 9 Jul 1999 16:56:56 -0700 (PDT)
          • Resent-Date: Fri, 9 Jul 1999 16:57:02 -0700
          • Resent-From: eskimo-announce@eskimo.com
          • Resent-Message-ID: <"W6n4H1.0.F91.DneXt"@mx1>
          • Resent-Sender: eskimo-announce-request@eskimo.com

          
          
               I just wanted to let people know that there are a number of hardware
          upgrades in the works and some downtime will be necessary on various
          machines. 
          
               We are upgrading eskimo, eskinews, isumataq, and chat.
          
               Eskimo is going to be moved from the existing 4/670MP to an SS-10. 
          This will provide a disk controller that is twice as fast as the current
          controller, an extra S-bus slot because in the SS-10 the 2nd M-bus slot
          does not overlap an S-bus slot so that you can use both, which will enable
          us to put additional controllers in if we needed to, and it will provide
          more bus I/O bandwidth, faster memory which will enable the CPU's to spend
          more time doing useful work and less in wait states. 
          
               Eskinews is also going to be moved from the existing 4/670MP to an
          SS-10, this will enable us to use all four CPU's instead of just two, will
          double the speed of one of the two disk controllers, and provide more bus
          bandwidth, all of which should be conducive to better news processing.  I
          am also talking to Skycache about a possible satellite news feed, pending
          a site survey that hopefully will resolve long standing problems with the
          feed from Sprint.  They are also upgrading their equipment to provide a
          45mb/s feed and that will provide capabilities like web and ftp cache
          pre-loading which ultimately could save us some bandwidth down the road. A
          full news feed if it were actually coming down the pipe at the full rate
          would eat up approximately 2.3mb/s of bandwidth which would fully saturate
          one T1 and part of a second.  So long term I really don't see any cost
          effective alternatives.  When we know what kind of feed rate we will
          really be getting we'll work on the disk issue.  There have been a number
          of new larger drives recently on the market that would enable an economic
          expansion of the spool.
          
               Chat is presently a 4/330 (25mhz VME based Sun-4 architecture
          machine) and is going to be replaced with an LX (50 Mhz S-bus based Sun-4m
          architecture).  The intent here is to load this new box with SparcLinux,
          setup SunOS emulation.  This will enable running existing binaries but it
          will also provide an ANSI compliant environment for compiling the newer
          eggdrop.  The LX is also version-8 Sparc which has hardware
          multiply/divide, so it is faster than the older Sparc even at the same
          clock speed. 
          
               Isumataq is a 4/260 presently, 16.7mhz VME based Sun-4 system.  It is
          also going to be replaced by an LX for pretty much the same reasons.  A
          lot of the newer game engines won't compile in a non-ANSI environment
          without a lot of work. 
          
               We are still trying to locate all of the necessary components for the
          file server, the one thing that seems problematic is finding a DEC FDDI
          card, DAS, for that machine. 
          
               I am also contemplating an upgrade of the mail server as the load has
          become quite heavy during certain hours of the day, a large portion of it
          being the result of spam related I/O.
          
               The upgrade of all of these machines is going to entail some
          downtime.  We will try to schedual it for low-usage hours but some
          downtime is unavoidable. 
          
          
          
          

          • Prev by Date: CGI's
          • Next by Date: News, DSL, and Meeting...
          • Prev by thread: News, DSL, and Meeting...
          • Next by thread: CGI's
          • Index(es):
            • Date
            • Thread