randybias
-
People conflate 'cloud', IaaS, PaaS, and SaaS value props. That's part of the definition confusion. These value props are different.
about 23 hours ago
from Tweetie
-
Is resource pooling really an attribute of 'cloud' or just IaaS? How does a PaaS pool resources? SaaS?
4:29 PM Nov 22nd
from Tweetie
-
@ Seriously there is no such thing as a tomato cloud! ;-)
3:53 PM Nov 22nd
from TweetDeck
in reply to Beaker
-
Will wonders never cease? @ decided IaaS == "Integration-as-a-Service' and they are 'leading' it.
3:52 PM Nov 22nd
from TweetDeck
-
@ @ homebrew looks cool, but no gimp...
2:40 PM Nov 22nd
from TweetDeck
-
@ What's a better replacement?
2:21 PM Nov 22nd
from TweetDeck
in reply to benjaminblack
-
RT @: Reading: Interview With Ezra Zygmuntowicz – Engine Yard [ nice job ez .... (cc@ezmobius) ]
12:48 PM Nov 22nd
from TweetDeck
-
First Snow Leopard problem. MacPorts did not just work. Looks like I have to rebuild or reinstall it. Bummer.
12:48 PM Nov 22nd
from TweetDeck
-
@ @ And I'm not really looking for a DC design. I'm exploring options for various architectures that tradeoff price 4 perf
7:53 PM Nov 21st
from TweetDeck
in reply to bradhedlund
-
@ @ 2 * 10GE or 6 * 1GE are fairly typical at the moment. High levels of redundancy aren't typically necessary.
7:53 PM Nov 21st
from TweetDeck
in reply to bradhedlund
-
@ 30 per box, one 10GE storage pipe per compute node.
4:34 PM Nov 21st
from TweetDeck
in reply to bradhedlund
-
@ Will it work? No idea. I'm just trying to understand what the real actual hard reqs on FCoE are as they relate to Ethernet.
4:34 PM Nov 21st
from TweetDeck
in reply to bensons
-
@ This switch should be able to do line rate with any drops. And it is not so-called 'lossless' ethernet. It's just overbuilt.
4:33 PM Nov 21st
from TweetDeck
in reply to bensons
-
@ Yes, I'm clear about that. I'm aware of a switch vendor bringing a switch to market with 100x the normal buffering.
4:33 PM Nov 21st
from TweetDeck
in reply to bensons
-
RT @: turns out, Rubinius has actors (not documented), and an experimental Multi VM (MVM) branch as well! [nice]
3:06 PM Nov 21st
from TweetDeck
-
@ I'd like every VM to be able to get at least 1 SATA disk of perf bursting to as much as 4xSATA disks
2:51 PM Nov 21st
from TweetDeck
in reply to bradhedlund
-
@ For me the problem is easy: backing store for VMs with maximal price/perf tradeoff and can be a 100% homogenous env
2:50 PM Nov 21st
from TweetDeck
in reply to bradhedlund
-
@ I'm not sure that finding a vendor first makes sense. I'd rather state the problem and a cost objective, then find a solution
2:49 PM Nov 21st
from TweetDeck
in reply to bradhedlund
-
@ If iSCSI is *not* equiv, then why and what are the tradeoffs? This all very honest questioning. I really want to know.
1:43 PM Nov 21st
from TweetDeck
in reply to bradhedlund
-
@ I'm trying to understand what the best path is for a greenfield DC. Why embrace extra complexity with FC if iSCSI is equiv?
1:43 PM Nov 21st
from TweetDeck
in reply to bradhedlund
|
- Name Randy Bias
- Location Stratospheric
- Web http://cloudscali...
- Bio Cloudscaler, CTO, ex-GoGrid, cloud computing, social networks, infrastructure, ZFS, ruby, UI design, mobile connectivity, poker, and general goodness
|
