Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!apple!metaphor!xymox!philf From: philf@xymox.com (Phil Fernandez) Newsgroups: comp.databases Subject: Re: TCP/IP over SNA (Sybase client/server) Message-ID: <772@metaphor.Metaphor.COM> Date: 16 Sep 89 23:17:10 GMT References: <72701@yale-celray.yale.UUCP> Sender: news@metaphor.Metaphor.COM Reply-To: philf@xymox.metaphor.com (Phil Fernandez) Distribution: na Organization: Metaphor Computer Systems, Mountain View, CA Lines: 53 Bcc: In article <72701@yale-celray.yale.UUCP> you write: >... >At our disposal is an SNA network connecting all the sites. We would prefer >to use it but if there are other cost effective ways we will certainly >consider them. The real questions here are whether TCP/IP over SNA is >possible/feasible/tolerable and whether Sybase DB-library functions well >over an internet I can give you a couple of answers here, from different experiences. I bought and installed Sybase for the main computing center at Stanford prior to recently re-joining the "real world" here at Metaphor Computer Systems. Stanford was running Sybase V4 on a Sun 4/280, supporting a bunch of clients around our campus TCP/IP Ethernet (SUNet). SUNet uses a backbone and spine topology, all 10Mbps, using primarily cisco Systems internetwork routers. Typically the clients were running on Sun 3/60 workstations. Some of our Sybase DB-Lib clients were 2 or 3 routers away from the Sybase server, and it all seemed to work pretty well. In this case, Sybase worked very well over the Stanford internet. I would easliy recommend production use in this environment. I'm less sure about using an SNA network as an intermediate. I can come at this from a different direction. I have no experience with TCP/IP thru an SNA net, but here at Metaphor, we have a product that performs XNS internetwork routing via an SNA network. we use this internetwork capability for a client/server model data base query system (but not Sybase). Our model is probably less "interactive" that a typical Sybase c/s application would be, but it still doesn't work very well. That is, we experience rotten performance across the SNA net. Typical SNA networks can have relatively long delay times (on the order of a couple of seconds), and this can be a problem with a highly interactive application. Our datagram routing through SNA is pretty optimized and is "state of the art" -- we route via an SNA LU6.2/PU2.1 connection, which is the best we're going to do at this time. So, I'd be careful... pmf +-----------------------------+----------------------------------------------+ | Phil Fernandez | philf@metaphor.com | | | ...!decwrl!metaphor!philf | | Metaphor Computer Systems |"Does the mind rule the body, or does the body| | Mountain View, CA | rule the mind? I dunno..." - Morrissey | +-----------------------------+----------------------------------------------+