LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: Turbo linux clusters as compared to lvs?

To: lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Subject: Re: Turbo linux clusters as compared to lvs?
From: wanger@xxxxxxxxxx
Date: Tue, 16 May 2000 11:30:13 -0400
On Tue, 16 May 2000 06:17:38 -0400 (EDT), Joseph Mack wrote: 

>You people read our mailing lists, have access to our code and yet never
>contribute anything here. Yet you are quick enough to pop up and defend
>your clean room reconstruction of our code, whenever it's mentioned. If
>you were really doing this independently of us, you wouldn't be on our
>mailing list. If you were on this mailing list in good faith, you would
>have contributed something by now.

Well, I jsut paid $995 for the 2-node version of TurboCluster 4.0.  
I've installed it, played with it, examined it, the LVS project has 
*nothing* to worry about.  We simply need a marketing machine to match 
the mouth of TurboLinux.  They are still using some incarnation of the 
old Red Hat 4.x installer (horribly broken), they couldn't support 
Compaq Proliant PII machines without me hacking my way though things 
(heck, their installer couldn't even detect a Mach64 video card).  Of 
course there is the fact that their network config tool supports dhcp 
client side, however, they don't ship the userspace bit (/sbin/dhclient)
that their scripts look for, etc. etc. etc.  Also notable is their very 
elegant method of hardware autodetection during the install.  Just 
insmod all driver modules and check the return codes for successes.  
Its a good thing they changed their name from Pacific Hi Tech as this 
stuff certainly isn't very "high tech".

Then there is the matter of their "fancy" GUI config tool for their 
clustering administration.  Well, they have a text based NEWT tool that 
supports modular config tool additions of which, they add 2 for cluster 
support.  Then when you launch, their "X" based tool, its the same 
thing, just using X fonts instead of the NEWT fonts and it runs in its 
own xterm.

Then we can talk about their kernel module, they have a significantly 
lower number of features of Wensong's code.  You know, things like NAT 
routing and least connections scheduling.  So, my conclusion is, we 
(the community) and companies (Red Hat, VA, etc.) that produce Linux 
clustering applications and tools have very little to fear.

Mike

-----------------------------------------------------------------------
Mike Wangsmo                                               Red Hat, Inc 

"I think qmail got mad, took its ball and went home." - Steve Wills




<Prev in Thread] Current Thread [Next in Thread>