John H. Robinson, IV wrote:
Neil Schneider wrote:
I'm interested in hearing more discussion about this.

Personally, I like the idea of our club maintaining our own server. It
gives us a flexibility in doing what we want to do, experimental or
community-wise.

The Xen and Xen-type hosting services are nice, as you get your own
root, but it can get expensive per month for the level of service
(bandwidth, storage and CPU) that we currently enjoy.

I agree that a 1U would be fine, with a couple 150G drives. We can even
RAID-1 them, for redundancy and still have copious amounts of space.

I know a local vendor that we can get a 1U with 2x150GB SATA drives,
SATA RAID controller, 2x XEON 2.8 (or is it 3.0?) GHz, with rails and
400 W power suppy for about 3,200$.

More complete details available if desired.

Looking in my latest issue of ComputorEdge, I see that SDcom <http://www.sdcom.com> has a 1U server with:

Intel Core2 Duo E6600 2.4GHz
1GB DDR2 667 RAM
250GB SATA II hard drive
Built-in Video, Audio, USB
1 Gigabit LAN
Floppy
DVD-ROM

Price: $1055

Expand storage to 2x 250GB RAID
Price: $1320

There are some other variations on this theme as well.

It looks like you can get a really decent server for under $2000. So this is bracketing the expense.

As Andrew already pointed out, if we want to implement full virtualization then we need to go with the newer chip. My preference lies that way also because it will handle future developments better. See <http://www.heise.de/english/newsticker/news/82344/from/rss09> for an article on KVM (Kernel Virtual Machine) that will make its appearance in the 2.6.20 kernel.

Of course we can wait a few more months to see what AMD will release. Apparently AMD is going to provide the ability to do hardware assisted Nested Paging. From a VMware article at <http://www.techworld.com/opsys/features/index.cfm?featureid=2957>

<quote>
He also showed a demonstration of a future AMD CPU feature, Nested Paging. This technology moves the mapping of virtual machine memory page tables to physical memory into the CPU hardware. Rosenblum demonstrated an extreme example of how applications could be accelerated -- AMD's Margaret Lewis said that acceleration of memory-intensive tasks such as compilation is likely to be around 43 per cent. Lewis said that this technology, first announced in August, would be available in 2007.
</quote>

Gus

--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-steer

Reply via email to