Article View: comp.sys.next.software
Article #73Re: Sybase 5.0?
From: Charles William Swiger
Date: Fri, 30 Oct 1992 09:33
Date: Fri, 30 Oct 1992 09:33
52 lines
3326 bytes
3326 bytes
Excerpts from netnews.comp.sys.next.software: 29-Oct-92 Re: Sybase 5.0? by andrew@cubetech.com > Porting to NetWare also makes much more technical sense. Sybase > properly written for NetWare will just blow the doors off a NeXT based > version and cheaper hardware to boot. Probably fare well against a > low end SS-10 if not being faster too. Hmm. Let me throw in some background for my comments. I worked over the summer with a company that produced software running off of SQL-based Relational Database Servers. They used Oracle & SQLbase, to be specific. They had a Compaq SystemPro '386 (33MHz, I think) & 16 MB of RAM, with a RAID drive array. (They had much faster machines around, '486-33's & 50's; but these couldn't use the RAID, so they continued to use the Compaq as the database server.) They got in a SPARCstation IPX to compare performance. Novell Netware versions were the fastest over either MS-DOS or OS/2 based versions of the RDS. The IPX (with 16MB of RAM) was around 30% to 50% faster than the SystemPro, meaning that a '486-based SystemPro would beat the IPX. A 32MB SPARC-2 was gotten, and it was substantially faster than the IPX, and would compete head-to-head with a fast '486. A SS-10 would blow away any '486 machine currently on the market for doing heavy calculations. A Turbo NeXT machine should fall into the performance range between the IPX and the SS-2. Note that the Netware-based systems do have very good transient response; ie. a simple query that can be answered by pages of the database currently in memory was dealt with very quickly. This seems to be related to the low overhead in network transmission for NetWare compared to the much higher overhead involved on a unix machine. But note that a '486 Compaq SystemPro is *not* cheap; in fact, low end SS-10's are cheaper. While there are much cheaper clones available, the company has extensively tested them in house, and examined customer installations using cheaper machines for a database server. They found out that the cheaper machines took a substantial performance loss, generally to unacceptable levels. This was primarily due to poor I/O throughput, both through the network card, to the hard drive(s), and even along the backplane of the machine (ie: some only had 16-bit memory buses). (PS: they've noted that Compaq has made substantially better machines for high-performance servers than IBM...) -Chuck +---------------------------------------------+ | Charles William Swiger -- CMU sucks rocks | "The television screen is the | AMS & normal mail: cs4w+@andrew.cmu.edu | retina of the mind's eye." | NeXTmail: <temporarily out of order> | +---------------------------------------------+ --Videodrome
Technical Details
Message-ID:
<AewMsZ600WBO83ryYy@andrew.cmu.edu>Path:
arc1-ca.pugleaf.net!sparky!uunet!ogicse!das-news.harvard.edu!cantaloupe.srv.cs.cmu.edu!crabapple.srv.cs.cmu.edu!andrew.cmu.edu!cs4w+Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!ogicse!das-news.harvard.edu!cantaloupe.srv.cs.cmu.edu!crabapple.srv.cs.cmu.edu!andrew.cmu.edu!cs4w+ From: cs4w+@andrew.cmu.edu (Charles William Swiger) Newsgroups: comp.sys.next.software Subject: Re: Sybase 5.0? Message-ID: <AewMsZ600WBO83ryYy@andrew.cmu.edu> Date: 30 Oct 92 09:33:57 GMT Article-I.D.: andrew.AewMsZ600WBO83ryYy Distribution: usa Organization: Senior, Chemistry, Carnegie Mellon, Pittsburgh, PA Lines: 52 In-Reply-To: <1992Oct29.010847.6090@cubetech.com>