Boston Linux & Unix (BLU) Home | Calendar | Mail Lists | List Archives | Desktop SIG | Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings
Linux Cafe | Meeting Notes | Blog | Linux Links | Bling | About BLU

BLU Discuss list archive


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

Ubuntu being used in Cisco cluster project...



On 8/20/07, Matt Shields <mattboston-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org> wrote:
> Finally got a chance to see the video.  Great video, I hope you win.

Thanks dude :-)

> Any chance you could provide a little more detail on what you're doing
> and how you do it?  I'm assuming your doing load testing using the 70
> host servers (4096 virtual servers). What are each of the vservers
> actually doing?  I don't think my company would do this scale, but on
> a smaller scale of 10 host servers they might.

So, for normal testing, we only utilize up to ~100 virtual machines.
This covers the various permutations that are possible with our
product.  For instance, if we build a new Cisco Security Agent
tomorrow, we want to run through our test/regression suite on all
supported OS releases.  This would include Windows NT4
Workstation/Server SP6/6a, Windows 2000 Professional/Server/Advanced
Server SP0/SP1/SP2/SP3/SP4, Windows XP Home/Professional SP0/SP1/SP2,
Windows Server 2003 Standard/Business/Enterprise SP0/SP1/SP2 +/- R2,
Windows Vista Business/Enterprise SP0/SP1.  So, from the web-frontend,
we pull down the management installer executable over Samba and run an
automated task using win32GUITest in Perl.  Then, when that finishes,
it reboots, and we install the client from the MC.  That will install
on all the supported platforms.  Then, we pull down an executable on
those to run through each test.  But this is all done however the
client automater wants to do it.

In terms of scalability, think about this.  Cisco Security Agent has a
new feature in version 6.x that will allow it to identify even the
propagation of mutating worms.  So, this algorithm requires 3 payloads
to create a signature.  So, say a bunch of machines are up, we let the
worm go, and after a few infections, the clients report back the worm
signature to the management console, the MC stores this information
and creates a unique filter, then redistributes this to all hosts in
the network.  What this means is that even brand new worms cannot
infect more than a few machines in your network, because it will be
identified very early and a signature will be distributed quickly...
-- 
Kristian Erik Hermansen

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.







BLU is a member of BostonUserGroups
BLU is a member of BostonUserGroups
We also thank MIT for the use of their facilities.

Valid HTML 4.01! Valid CSS!



Boston Linux & Unix / webmaster@blu.org