Sunday Sep 27, 2009
Word of mouth is the best marketing. Literally in this case. My neighbor told me another new feature was delivered to AT&T set top boxes to allow you to play media content from your hope PC's to your TV. Already of course I could sync my iPod, and carry it down to the docking station in the living room. Or I could write music and photos to a USB stick and plug it into the TV to see and hear it. Or I could upload content to Yahoo and view it through U-verse.
This is different. The set top boxes simply connect to Windows Media Player on PC's through the home router to access content without sending anything outside the home. Well this won't work for me, I thought, because I don't run Windows native. I run Windows XP under VirtualBox virtualization software.
I only had to install Media Player 11 (version 9 is too old) and select "bridged adapter" networking for my Windows virtual machine. (I doubt that NAT would work since the virtual machine wouldn't be visible on the network, but I didn't try it.) Now all my TV's can browse music and photos on two PC's. Select background music and start a slide show of today's snapshots. Really nice.
Wednesday May 27, 2009
Another
SPECtacular award from the SPEC annual meeting: Paula Smith
(VMware) was honored for her
tireless, competent and patient work managing the SPEC office and the
people there. Paula consistently exhibits what make SPEC an unique
place. The attention and enthusiasm she brings to her volunteer work
make her a pleasure to interact with. She goes above and beyond in
everything she does, and is often able to turn emergencies into
opportunities. Most impressive is how she maintains this over time
and in every interaction, despite many competing pressures for her
attention. Beyond this management work, she also manages to handle
the organizational and technical work of chairing the Virtualization
committee, and of course her day job at VMware.
Thursday May 14, 2009
More 2009 SPECtacular awards. SPEC's
forthcoming virtualization
benchmark will provide meaningful metrics of hardware and
software performance in data center consolidation. As complex as this
benchmark is, running several different benchmarks together in
virtual machines on a host system under test, the code is only half
the story. As with all benchmarks the workload is vital, to represent
realistic usage scenario(s) so that performance improvements made on
the benchmark will also benefit real world users. And the run rules
are vital, needing to accommodate technology improvements over the
lifetime of the benchmark, while precluding unrepresentative
optimizations exploiting rule loopholes. (Or what the layman might
call “cheating”) There is spirited debate from companies
representing rather diverse user communities, all with an interest in
seeing that their customers' needs are addressed by the benchmark. In
the end when this group of top engineers reaches a consensus you know
they've come up with a benchmark that is as rock solid as is possible
to make. From among this great
team of partners and competitors, three were singled out for
SPECtacular awards:
Andrew
Bond of HP always steps forward when
a person is needed to test new code, features, parameter tuning. He
performed many experiments whose results showed the committee the
sensitivity of the benchmark to various parameters, sizes, and
configuration options, so that the right choices could be made for
fair benchmark comparisons. He also created scripts to set up and
configure new guest VMs for each workload.
Chris
Floyd of IBM improved and tailored
the mail server and application server workloads for the new
benchmark. He's revamped these workloads several times to improve
the I/O profiles and add burstiness to the application server
transaction injection. He helps the other developers at regular
on-line coding sessions, explaining new features, and resolving
problems. He even helps out when on vacation.
Greg
Kopczynski of VMware
developed a (necessarily) complex and feature extensive
harness for the benchmark. He responds to countless pleas for help,
assistance, debugging, etc., in true SPEC fashion without asking
whether the help is for a partner or a competitor. He added
burstiness to the web server workload. And he integrates new code and
changes from all the developers for each development kit revision.
Thanks for your great efforts!
Tuesday Feb 19, 2008
Virtualization is such a hot technology that Dilbert is poking fun at it: 2/12, 2/13 and 2/14. No wonder, since IT centers must use both hardware and energy more efficiently. At SPEC's 2008 annual meeting in San Francisco SPECtacular awards were given to members of the Virtualization committee. As always, I won't post anyone's name without permission, but you know who you are and SPEC is grateful for your contributions.
SPEC is working on a benchmark to model server consolidation of commonly virtualized systems such as mail servers, database servers, application servers, web servers, and file servers. Requiring a very different technical approach than SPEC's traditional benchmarks, virtualization has brought unique challenges. SPEC recognizes these engineers for outstanding contributions in meeting those challenges:
- Andrew Bond, HP
- Cathy Reddy, Unisys
- Chris Floyd, IBM
- Fred Abounador, AMD
- Greg Kopczynski and another engineer, VMware
- Nitin Ramannavar, Sun
- Stephen Pratt, Communigate
- and an engineer from a company so modest that they don't even want to accept public thanks.
Thursday Jan 10, 2008
Electrical, not political. A DOE study found that - duh - if you give
consumers information about time varying cost of electricity they will
save money by shifting some power usage from peak to off-peak times.
Consumers in the study lowered their electric bills by 10% and lowered
their peak demand by 15%. This is a big deal because although the
operating cost component of electricity (fuel) depends on the total
energy consumed, the capital cost component (generating plants) depends
on the peak power generation.
Solar power is particularly valuable to a utility because its peak
production occurs in the middle of the day when summer demand from air
conditioning is highest. But there's another peak around 5-6 when people
come home from work and turn on appliances, and by then solar power
production has fallen off. Thus adding photovoltaic power alone may not
drastically reduce peak requirements for fossil fuel power plants.
Wind power along the California coast has an almost complementary
generation curve to that of solar power, because of the onshore and
offshore breezes in the mornings and evenings. Adding wind power alone
may not drastically reduce peak fossil demand because the wind often
dies down mid-day when the air conditioning load is highest.
But adding solar and wind power together could greatly reduce peak
fossil demand, though perhaps not economically eliminate it entirely.
Then if you added time of day metering to allow consumers to voluntarily
shift their load, that would level even more peaks. Ditto various energy
storage systems like the plan to use night time wind power to pump water
back up a hydroelectric dam for use the next day, super capacitors, and
plug-in hybrid cars. The key to effective and economical use of
renewable energy is a balance of power supply with demand.
The computer industry tries to do the same thing with servers. Demand
for computing services typically follows daily, weekly, and monthly
cycles. When the data center is provisioned for the highest possible
demand, there is a lot of wasteful excess capacity. Even with the most
efficient hardware and the best power management software, running
servers at low utilization is extremely wasteful compared to moderate
utilization. So we try to balance computing supply with demand by
virtualization and workload consolidation, especially if we can find
workloads that are complementary (like wind and solar) in their resource
requirements and/or their load versus time of day.
As network capacities increase and software becomes more sophisticated, you can imagine systems configuring computing resources worldwide to
maximize computing power to the customer at minimum electric cost. Think
of a customer connected from California in the middle of a hot day with
time-of-day electric meters set to the highest price. Of course he might
be routed to servers in Europe or India where the computing demand is
off peak. He might also be routed to servers in Colorado where the
computing demand might still be high, but the electricity demand and
price might be lower. Or to Oregon where a heavy rainfall and cold wave
might mean cheap renewable hydro-power, even at peak electric demand;
and lower than usual data center cooling costs thanks to mixing filtered
outside air.
Wednesday Oct 31, 2007
"Is storage becoming IT's Hummer?" asks The Register. Reporting from SNW Europe they report that as data centers reduce the power cost of computing, storage is poised to become the biggest energy consumer. Well that's just the outcome SNIA hopes to avoid with their Green Storage Initiative. The efficiency race between computing and storage is one where we can cheer for both sides. Besides, as Jonathan points out, the distinction between computers and storage is blurring fast.
The Reg says that virtualization will be primarily responsible for reducing computing power usage through consolidation. Certainly the most effective way to save energy is to follow your mother's command: "Turn that thing off if you're not using it!" But I think the Reg is a bit premature in giving the industry credit for solving the problems of computing power usage. Yes there's a lot of innovation in this area making data centers more efficient in many different ways. But there remains a lot of hard work to do by vendors and users alike.
|
|
|
|
