| WIKI | FAQ | Tech FAQ | |
 http://csua.com/feed/ | |
| ||||||
| 2008/11/29-12/4 [Computer/SW/Languages/Java] UID:52127 Activity:nil |
11/28 python really is going hafter the java market aren't they:
PEP 3119: Abstract Base Classes (ABCs); @abstractmethod and
@abstractproperty decorators; collection ABCs.
this is after guido saying "we do not need abstract base
classes/interfaces" for how many years? |
| 2008/11/29-12/4 [Computer/Networking] UID:52128 Activity:moderate |
11/28 So you know, I am going to see if the collective motd.wisdom has
anything to add to my own research. I would think that given my
political leanings in most areas, I would be a big fan of "net
neutrality" but I am not, at least not so far. What is wrong with
someone like AT&T charging more for premium internet service? Don't
they do that already? Is there something I am missing here? -ausman
\_ My understanding of net neutrality is to not prefer one packet type
over another. Paying more for higher service seems reasonable, but
(say) dropping VOIP packets so Skype doesn't compete with AT&T phone
service seems a bit sub-optimal.
\_ Right, premium services are fine. The worry is abuse of
monopoly power -- that's why net neutrality is focused on
last-mile providers (which often have significant monopoly
power), rather than backbone providers (which have less).
A backbone provider could never get away with blocking VOIP,
because their customers would just switch to a different
provider; but your local phone company might be able to, if
they control the only point of access to your house. Given
sufficient monopoly power, ISPs might even be able to engage
in more blatant extortion, by (say) threatening to block
their customers' access to your website unless you pay them.
\_ Actually that was how net neutrality got to be such a big
deal. One of the backbone ceos, I forget which, basically
started hinting that he'd like to make it so unless google
started paying protection money they were going to get
lower priority packets.
\_ Any source for this? I know Google has waived a red flag,
claiming to be worried about this, but I have not seen
anything from a backbone provider. The current powers on
the Internet like things just fine the way they are, and
have an obvious interest in stifling innovation. -ausman
\_ I thought it was from the Comcast CEO, but it seems the
original quote was from AT&T CEO Ed Whitacre circa Nov 05
"Why should [Yahoo, Google] be allowed to use my pipes?"
http://preview.tinyurl.com/9svdw
http://preview.tinyurl.com/6ooh2u
\_ Thanks for the pointers. I was not aware of this.
\_ Thanks for the pointers. I was not aware of this. -a
\- transmission to ausman: i havent thought about this deeply but
my inclination is NET NEUTRALITY is a form of PRICE DISCRIMINATION
which leads to CONSUMER SURPLUS being shifted to the firm.
for background, a very, very good book is JTIROLE: IO. I dont
have the inclination to type on this more, but you can google
around ... it looks like there are some others who buy this
approach. --psb
\- i thought about a little more in the background, and other issue
is these narrow arguments in terms of efficiency may ignore some
distributional consequences ... same as say congestion pricing,
efficient pricing for publicly owned parking etc. --psb
\_ There's more to it than you are seeing. It's not just about your ISP
charging you for different tiers of service. They already do that
now. It's about the ISP then turning aruond and trying to charge
various services for delivering content to you. If they don't pay
up you get crappy performance to those sites/services. It is all
about the ISP trying to double-dip on the income stream. It also
very much kills the 'openness' of the internet. Would you still
want service from an ISP that only granted you access to the sites
that ponied up their non-neutrality fees? What if you had no choice
in ISP any more?
\_ Why would you get any crappier service to those sites than you
already do now? If I didn't like my ISPs traffic shaping policies,
I would just switch ISPs, right, just like I can do now? This
could potentially be a problem where there is a monopoly on last
mile service, but that is getting rarer and rarer, with cable
and satellite available. I might want to pay extra for things
like HD quality streaming video, which I can't do now. Why
is that a problem? I just don't see this as killing anything,
other than perhaps some big Internet companies profit margins,
hence their scare tactic lobbying campaign. -ausman
already do now? If I didn't like my ISPs traffic shaping
policies, I would just switch ISPs, right, just like I can do
now? This could potentially be a problem where there is a
monopoly on last mile service, but that is getting rarer and
rarer, with cable and satellite available. I might want to pay
extra for things like HD quality streaming video, which I can't
do now. Why is that a problem? I just don't see this as killing
anything, other than perhaps some big Internet companies profit
margins, hence their scare tactic lobbying campaign. -ausman
\_ again, you're looking at it frmo the standpoint that
a) you have choice in ISP's, many don't.
b) you have a pipe to an unfettered internet, which is what
we have now with neutrality. With non-neutrality, you'll
see degraded (or nonexistant) services to places on the
internet who don't pay up. And the only way to find out
your ISP's qos to a given site would be to go there. Good
luck shopping around other ISP's to see who's got the best
performance to your favorite sites.
c) its not going to hurt the big internet companies much,
they take some minor hit to their bottom line and pass
on the costs. Its the small and up-and-coming companies
that will get hurt, as this 'pay to play no the ISP's nets'
tax will just add to their barrier to entry. Taking away
neutrality would be a step in a return to the balkanized
network days of compu$erve.
network days of compu$erve and prodigy.
\_ Except you have it exactly backwards, in that currently
network owners can do as they like and the only
pressure on them is their customers. Proponents of
"Net Neutrality" want to pass laws that freeze the
status quo in legislature. If blocking small up and
coming sites were a problem, wouldn't ISPs be doing
that all ready? Do you really want Congress determining
which packets and protocols should be used on the
Internet, instead of the IETF?
\_ Um no they can't. Right now they do 'best effort'
to everywhere, with no discrimination against
sites who haven't ponied up good-performance
extortion fees. No we dont want congress
determining which packets/protocols should be used
-- they dont do that now. That is the status quo
we want preserved. When comcast tried to disrupt
bittorrent traffic, and the FCC started
investigating, they backed down.
\_ Why can't they? People set up spam filters,
firewalls, etc all the time, why couldn't an
ISP do the same? There is no law against it,
though the NN folks seem to want there to be
one. Am I right about this? Link to the FCC/
Comcast situation please? Using the FCC to
stop technological change doesn't exactly
strengthen your case, btw.
one. Am I right about this? The FCC/Comcast
case is currently pending on appeal, btw. |
| 2008/11/29-12/6 [Computer/SW/OS/FreeBSD, Computer/SW/OS/VM] UID:52129 Activity:moderate |
11/29 I'm experimenting with virtualization, and as a poor college student
I'm wondering what the best alternatives for virtualization are, and
how best to cut my teeth on messing with non-linux platforms (or I
guess interesting stuff on Linux would work too). Right now I've got
FreeBSD7 running on KVM on my home computer (on a Core 2 Quad), and am
somewhat at a loss as to how to use it. (More details: bridged
networking, disk is a 8GB partition software raid1'ed over 3 disks).
In any case, KVM seems to just 'work', but as the CSUA is planning to
offer VMs soon, I'd like to know if there are better alternatives,
particularly considering that when I put my computer to sleep without
shutting down the guest OS, the computer wouldn't start back up, I had
to cold-boot, and the disk image got corrupted. From what I hear,
VMWare's offering is solid, but the useful administration software is
thousands of dollars. Ideally, free software or something sustainable
without repeated donations of software, and easy to administrate would
be best. Does anyone have suggestions? --toulouse
\_ At my job, we use Vmware 2.0. it is free. i run vms. there
are graphical admin tools. I could buy Vmware ESX, which gets
me I guess better admin tools, better performance vmotion and fail
over.
\_ Someone here works at VMWare and was recruiting 2 years ago.
Calling the VMWare guy! We need a free educational license!
Oh well, he's probably not going to respond until Monday.
Us old farts have kids and family things to go on weekends.
Oh, try this. And yes we use VMWare in our company and it
is really great. You can get snapshots of the machine, run
multiple instances on a single machine (since most machines
are underutilized). Our production servers are also in
VMWare for superior bug isolation and debuggability:
http://www.vmware.com/partners/academic
\_ What, you mean CSUA alums have lives? Unthinkable! --toulouse
\_ Isn't VMWare Server free? That's what we use in our company.
--- !OP
\_ I don't recall the details, but while the server itself is
free, I think the administration interface is expensive.
Feel free to correct me on this. --toulouse
\_ Here's the deal. Vmware has two products. The Free Version
(Vmware 2.0) , and Vmware Server ESX ( not free. lots of $$$$ ).
ESX is a different codebase than Vmware 2.0 free. With ESX,
you get better performance, better GUI tools, failover capability,
and the ability to magically move your VMs from machine to machine.\
freely available Vmware 2.0 has a gui too.
and the ability to magically move your VMs from machine to machine.
freely available Vmware 2.0 has a gui too.
\_ VirtualBox?
\_ virtualbox is a sun thing. its not vmware. it
has its strengths and weaknesses
\_ ESXi, the hypervisor, is actually free, it seems, but the magical
admin tools are a part of ESX and not ESXi:
http://www.vmware.com/products/esxi
Anyways. Paging VMWare employees...anyone here?
--Andy
Anyways. Paging VMWare employees...anyone here? --toulouse
\_ dude are you running a root name server? Vmware 2.0 is
just fine.
\_ doesn't mean ESX wouldn't be better ;). Ease of admin is a
real concern for us, and besides, if the software is
satisfactory, we might even virtualize soda itself. Given
time, if we got another server with virtualization
extensions, failover would be a large win. As you may have
noticed from recent downtime, Keg's been on the fritz
lately, so uptime's been on our minds. Without failover,
we're back to square one re: evaluating KVM vs VMWare vs
others, hence this thread. Besides, there's an argument to
be made that if we have experience managing the good stuff
here in college it'll be what we're qualified to manage once
we strike out in the real world, and/or the software that we
recommend to our superiors should we get relevant jobs
(which, arguably, a few of us will). --toulouse
here in college it'll be what we're qualified to manage
once we strike out in the real world, and/or the software
that we recommend to our superiors should we get relevant
jobs (which, arguably, a few of us will). --toulouse
\_ I guess. Really, I think Vmware 2.0 is adequate.
There are plenty of cheapass companies out there running
it.
\_ You know, when I was a poor college student, I
wasn't very picky. Seriously, the two may have
different features that you'd need in the enterprise
environment, but are you running an enterprise?
\_ Well, I'm not picky wrt/ using what works for me
(which, as I mentioned before, is KVM), but I
want the CSUA to be a bit more ambitious in its
endeavors, and as they say, shoot high, aim low
(is that the right saying?). Plus, there's the
fact that our vp is not paid, so minimizing the
addition to his workload while offering more
students to members is also a factor. In any
case, I think it'd be prudent for us to see if a
software donation is feasible, and if not, what
our other options are then. This is something
that can wait a bit, as we're waiting on those
core i7's. --toulouse
\- (80cols ... reformatted)
\_ well if this is about the CSUA rather than
personal edification, how about first dealing
with the frequent crashes/outages of soda ...
or is this an attempt to do so? [this seems
odd to me, but whatever]. second, to abuse
a quote a bit, "software is the continuation of
policy by other means" ... "what [csua] problem
are you trying to solve" [via this software, via
donation campaign/new hardware etc]. BTW, with
regard to giving csua people experience with
expensive tools, i actually think part of the
reason a lot of ex-csua people have been
successful systems people is they resorted
to hacking togethe things and thus understanding
how they work under the hood, rather than
throwing money at the problem [hardware and
softwarewise] ... i'm not saying you should say
solve all problems that way ... like if you need
disk space today, just go buy a cheap disk
rather than scrounging, but just the observation
in the past, some of this hacking to debug
something or getting it to work (and much of
this was pre-google) served people well.
\_ Yep, real learning comes as part of the
struggle. In some sense, it would be better
for students not to primarily have experience
with enterprise software packages since
these are made "easy to use" for the corporate
drones who wouldn't survive if they had to
have any real degree of understanding of how
the system actually works.
\_ Well, the learning I was looking for when
putting the idea forward (since I suggested
it) was geared towards people exposing
themselves to different OS'es and playing with
root in a sandbox. This is the problem I want
to solve, not training people in enterprise
applications. Also, soda hasn't been crashing
-- it's been keg, which serves our LDAP, that
(as I said before) has been on the fritz. If
keg goes down, then logging in does not work.
Politburo intends to buy a new server for
this; however since the Core i7 is coming out
we don't want a purchase now to be obsolete
upon arrival. We have the opportunity now to
solve two problems at once: allow interested
members access to their own personal VMs, and
increase stability of our servers. We can most
definitely do without failover, but then the
uptime problem isn't as completely solved.
The idea of getting students experienced in
adminning VMWare may be of low priority for
the CSUA as a whole; on the other hand, it is
(IMO) the strongest argument to be made to
VMWare.
In summary (and in my opinion) -- high
priorities are increasing uptime and developing
skills with adminning systems.
low priorities are developing VMWare admin
skills and...well, steven should be coming on
soon to offer his opinion. --toulouse
\_ Can't you request a free license for VI3 from VMware at
http://www.vmware.com/partners/academic
\_ You could try virtual box from Sun, it is free and runs many x86
OSes:
http://www.virtualbox.org
http://www.virtualbox.org/wiki/Guest_OSes
Re VMWare - Fusion for OSX is very nice and quite affordable (I've
seen it on sale recently for as little as $30). It has GUI admin
tools and the unified mode makes using windows apps almost like
using native OSX apps.
I'm currently using Fusion to run WinXP and Ubuntu and have used
it in the past to run Solaris x86 and FreeBSD as well. I usually
run XP and OSX concurrently and haven't ever had any problems with
the XP VM getting corrupted when I sleep my iMac. If you have a Mac
I'd recommend getting it.
\_ I don't think you understand what he's trying to do.
\_ Maybe I misunderstood, but isn't part of what he is trying to
do is becoming more familiar with non-linux systems ("I'm
wondering ... how best to cut my teeth on messing with non-
linux platforms"). If he has a mac, Fusion is a good way to
accomplish this - it can run Solaris, Linux, *BSD, Windows,
&c. and will help him get a feel for those systems. Virtual
Box, while not as nice as Fusion (at least on Mac), is a
free way to accomplish the same.
\_ These are two different objectives. I'm talking about setting
up VMs as a service for CSUA so we can consolidate our machines
while maintaining some sort of security and OS diversity (linux
+ BSD at least) If toulouse wants to learn about
virtualization
of course Fusion is a good option (he does have a mac), but
that's a different aim. --Steven
up VMs as a service for CSUA so we can consolidate our
machines while maintaining some sort of security and OS
diversity (linux + BSD at least) If toulouse wants to learn
about virtualization of course Fusion is a good option (he
does have a mac), but that's a different aim. --Steven
\_ Hey guys - Steven here
Thought I'd weigh in on the situation. The recent outages have
indeed been because Keg has been crashing (as presumably toulosue
pointed out) and I'm fairly sure it's a hardware issue. We're
simply running too much IO through the (decently old) system
and parts of it have already failed (we've lost one of the
ethernet controllers already) so I'm willing to blame the system
instead of the software. That said, we're hoping to buy a massively
cool system when Core i7 Xeons come out (thinking 16+ cores). At
that point it seems reasonable to look at virtualization. I've
used Fusion and Virtualbox in the past, so I'm not new to it
by any means - but one of the requirements is that it's easy to
admin/use. The issue here is the host OS - I'd like to use ZFS for
the disk array we'd need to have to back all this. Linux doesn't
seem to have a very good filesystem for this sort of thing - ext4
isn't stable, btrfs is still even further off, ext3/LVM is pretty
hacky, JFS/XFS really really need battery backups to not lose data,
and reiserfs's future is very unstable.
ZFS offers ZVOLs which seem to be perfect for giving out virtual
partitions. Right now we have Soda mounting off of Keg via NFS
which as you may have noticed is a serious performance and stability
problem, so I'd prefer not to go with NFS again. The network FSes
out there all seem to suck in one way or another, so local storage
(especially for something like this) seems to be a must.
Since that limits us to using FreeBSD or OpenSolaris as a host OS
\_ or OSX, see:
http://preview.tinyurl.com/5zo987
[developer.apple.com - zfs(8)]
\_ We're not rich enough to buy a
Mac Pro/XServe :(
(unless Linux ends up having a decent fs by the time we actually
get this running). Virtualbox doesn't seem to work well on FreeBSD
(as in not at all) and Xen seems to not play nicely with either
BSD or Solaris as a dom0. VMWare won't run on BSD either - not
sure about Solaris, which is why I was looking at ESX. The problem
with ESX is that it runs on only about 3 supported hardware
configurations which are pretty hard to build on our budget.
Discuss?
I'll hang around and maybe get into this whole motd thing ;)
\_ Virtual Box on OpenSolaris w/ ZFS sounds like it would probably
work. I used to know some OpenSolaris people when I was at sun,
and could probably put you in touch with them if you run into
problems. -ex-Sun
\_ That'd be neat, I'll do so if we go that route and have
troubles |