|
Softpanorama
(slightly skeptical)
Open Source Software Educational Society |
May the
source be with you,
but remember the KISS principle ;-)
|
Softpanorama University
(Slightly Skeptical) Solaris Page
Home Related
About
Recent changes:
Solaris 10 is the first Unix that deserves to be called "XXI century Unix" due
to implementation of light-weight virtualization scheme called zones, extension
of RBAC and process right management, DTrace, ZFS and predictive self-healing.
Those distinctive features mean that Unix as the whole (including but not limited
to multiple BSD and linux flavors) will continue to move forward with other Unixes
coping and moving further the best Solaris 10 features.
| Solaris 10 zones implementation (which includes
Solaris-specific seamless integration with RBAC and process right management),
is one of the most important innovation in Unix for the whole history
of existence of this OS |
Another nice thing is Sun partnership with AMD and getting Solaris 10 on Opteron
on equal footing with Solaris 10 on Sparc. Sun enjoyed a vert good timing on Sun's
part to capitalize on Opteron architecture, which eventually (and not without Microsoft
help :-) became standard for Intel too and which includes important innovation
such as 64-bit and virtualization suppport.
While Sun marketing slogan about Solaris 10 "ten steps ahead " of competition
is probably somewhat of a stretch I would say that the following four features really
distinguish it from competition:
- Zones are light-weight virtual
machines and as such provide new and unique functionality including dramatic
improvement in Unix security. As overheard is in single percentage points,
zones can be used to isolate important applications and processes with each
instance maintaining own TCP-IP stack, the set of supplementary applications,
set of users, and what is the most important its own root account (avoiding
turf battles with application developers that compromise Unix security more
then anything else in any large organization). This is a completely different
solution from VMware which along with its classic predecessor IBM VM/CMS are
heavy full hardware emulators with each instance running full OS. They were
inspired by FreeBSD Jails. Like any light weight VMs they share memory and kernel.
But the level of separation of filesystem is pretty high: the only way to communicate
with an application inside a zone is via network services like NFS. I
would like to stresses that this method of isolating applications from each
other and from "mother ship" is a new Unix security paradigm that is pretty
natural to use for all but the most convoluted applications (I would not recommend
running Oracle in a zone if you still have some hair on your head; at least
not right now ;-).This is the most existing feature of Solaris 10:
- Due to low overhead, one server could have several dozens of zones and
still be able to perform some useful work :-)
- If a service in the zone is compromised, the activities of the attacker
will be constrained to the zone, but also will be fully visible to
the administrator, at minimal risk to the administrator. This model offers
substantially enhanced monitoring in comparison with full virtual machines
like AIX LPARs, that offer little reliable insight into their operation
once compromised.
- Constraints on system calls greatly hamper the ability of the attacker
in employing Rootkit.
RBAC implementation and
Process Rights Management
allows users to be granted full rights for specific processes, making root
access unnecessary for all but a few administrators.
DTrace: DTrace is new tracing facility built into Solaris. It improved
the ability to identify system problems and bottlenecks. It is a huge system
and the Solaris kernel instrumentation for achieving this functionality is an
achievement in itself ( it has a 41-chapter manual). The tool is script language
controlled. It uses D - the DTrace language which is similar to AWK. No
other OS is currently even close to Solaris 10 in this area.
Predictive self-healing: I always hated typical corporate "high availability
solutions" with the devices like F5 and/or clusters. F5 style redundancy
as typically implemented is an expensive way to introduce more single points
of failure. IMHO the current F5 installations mainly provide an expensive way
to shut down one box for maintenance while the second (partially out of sync
if there is no common SANs storage with the automatic switch) box imitates on-line
activity when nobody probably cares. This is also a nice way to learn something
new about problems of abusing TCP/IP (subtle data corruption of TCP streams)
and the more complex protocol is, the less reliably devices like F5 handles
it.
It looks like Sun heard my prayers. Predictive Self-Healing is a newly
designed hardware architecture that has capability to be reconfigured dynamically
by software if diagnostic software determines that a component started to go
south. Currently it can switch off faulty CPU, banks of memory and power supplies
(of course if there are redundant power supplies on a particular server; but
if you got them they are hot-swappable). Disks were not a problem for
a long time. Power supplies diagnostics is still buggy in some v210 probably
due to faulty temperature sensor, but generally works OK. May it Solaris
can switch off something else too :-) but just CPU and memory are probably
enough to provide 95% of a typical "high availability solution" benefits without
typical "high availability" overhead and complexity. That means that a slight
premium that you have to pay for UltraSparc servers is now more then justified,
especially for large corporate environment were cost of the server is always
more then offset by overhead inherent in running any huge organization...
ZFS: ZFS (which still is raw and as of late 2006 probably needs another
year to stabilize) eventually might became the best among the most modern Unix
file systems. ZFS makes a volume manager redundant and has many other
interesting features.
Meanwhile I would like a simple addition of BSD style attributes to classic
UFS which remains tried and true filesystem for conservative users :-).
Actually Sun made UFS logging default in Solaris 10 (it was available in Solaris
9 too, but not as a default -- you need explicitly put keyword logging into
dfstab for each partition where you want it -- and that means that it largely
ignored by system administration masses).
Dr. Nikolai Bezroukov
Notes:
- Those pages are written by people for whom English is not a
native language. Some amount of grammar and spelling errors
should be expected.
- This is a Spartan WHYFF (We Help You For Free) site. It
cannot replace the best teachers and
the
best books.
- The site contain some obsolete pages as it develops like a
living tree... Some links on older pages
are broken. Please
try to use Google, Open directory, etc. to find a replacement link
(see
HOWTO search the WEB for details).
We would appreciate if you can
mail us a correct link.
|
|
About: Patch Check Advanced (pca) generates lists of
installed and missing patches for Sun Solaris systems and optionally
downloads patches. It resolves dependencies between patches and
installs them in the correct order. It works on all versions of
Solaris and on both SPARC and x86.
Changes: HTML tags in patchdiag.xref are ignored. This
change from Sun to patchdiag.xref breaks compatibility with all
previous versions of PCA and makes updating mandatory. An option for
concurrent patch downloads was added. A new option to set sunsolve
access protocol to HTTPS was added. wgetproxy options for non-SunSolve
URLs are honored as well. The file ../etc/pca-proxy.conf is read in
proxy mode. Checks for several patches were added.
Cloud Computing with Joyent Accelerators
Software previously installed on personal computers is being
shifted or extended to be accessible via the Internet. The term for
this is “cloud computing”. In the cloud, you run your application on
one (or many) virtual computers. The advantages to this approach
include reduced up front hardware costs, reduced maintenance and
administrative costs, the ability to match hardware and software
capabilities and, most importantly, the flexibility to scale your
site up and down depending on demand.
Joyent’s Cloud
Our Accelerator™ powered compute cloud provides a highly scalable
on-demand infrastructure for running web sites, including rich web
applications written in Ruby on Rails, PHP,
Python and Java. Joyent Accelerators are next-generation virtual
computers that can grow and multiply (or shrink and consolidate)
depending on the real world demands faced by your Web application.
Accelerators are built on OpenSolaris, multi-core (8+),
RAM-rich servers (32GB+ each) and vast
amounts of NAS storage. Accelerators are
deployed in the best routing and switching fabric (Force 10) and the
best load-balancers (F5 Networks) available (and always will be).
Proven Track Record
- We have the largest OpenSolaris installation in the world
- Joyent manages hundreds of TBs of storage in each datacenter
- We serve billions of web pages each month for our customers
- Joyent delivers and receives hundreds of millions of emails
every month
Joyent’s Storage is Real, Not Virtual
OpenSolaris’ ZFS filesystem provides
seamless access to as much NAS powered
storage as you need. This means that if the system goes down, you do
not need to suffer through a long boot as you re-instantiate an
entire virtual machine, with data, from a remote data store.
Instead, your data is still there on your disks and the Open Solaris
means you can be up and running within 10 seconds.
Rails Applications Rock On Joyent
Joyent has a long successful history of scaling Ruby on Rails Web
applications to thousands and tens of thousands of requests/second,
including our own Joyent
Connector collaboration suite. If you build in Rails, Joyent’s
experience can prove invaluable.
Open and Standards Based Means No Lock-In
We are going to try and get you hooked on our great support,
first class infrastructure and flexibility; but we are never going
to lock you in. If you become the next billion dollar Web wonder,
this means your set-up is entirely portable. Of course, you then
have to invest a huge amount in off the shelf hardware.
Buy what you need, when you need it
Great post ! "What irks me is that many in the Linux community seem to *want*
Sun to fail. This is discouraging and totally counter-productive to the ideals of
Free Software that most in the Linux community claim to adhere to."
April 28, 2008 |
Safe as MilkI’ve been annoyed by some of the Sun-bashing that has been
going on over the past few months and years. I’ve
written
in the past about my belief that Sun are trying to do the right thing, and
my appreciation for the investment that they’ve put into projects I care about.
And yet no matter what they do, it seems like there are nay-sayers working to
undermine Sun’s community-building efforts at every turn.
Here’s a few examples of Sun-bashing that I’ve seen recently:
- No projects primarily sponsored by Sun get accepted to the
Google Summer of Code (unless you count MySQL). Rumour has it that Sun
were told not to bother applying. Of course the Summer of Code is Google’s
baby, and as such they decide who gets to participate and who doesn’t. They
don’t even have to explain themselves.
- Linux Foundation
employees
repeatedly criticising OpenSolaris and Sun. I suppose that this is to
be expected from a group that is representing its members, and sees the
OpenSolaris kernel as direct competition to the Linux kernel, but it’s just
as disappointing to me as when I see KDE
or GNOME hackers ripping into each other
- Press articles in
Slashdot
[2] [3]
and elsewhere consistently spinning things as “Sun’s free software efforts
aren’t sincere” interspersed with “Sun is ruining <insert project here>”.
I feel like a lot of this rhetoric is self-fulfilling prophecy. If you say
often enough “Sun is a bad community player”, then Sun’s projects will seem
unattractive to prospective volunteers.
All of this completely ignores the many great free software people who are
working for Sun - to name just a few, Glynn Foster, Simon Phipps, Dalibor Topic,
Ian Murdoch, Rich Burridge. These people are extremely clueful about free software
and community interests. And the message which we have seen consistently from
Jonathan Schwarz over the past couple of years reinforces that there is a commitment
to free, community developed software, and there are many capable people working
towards that commitment within Sun.
So why the difficulties? Many of them, I think, are project specific, and
stem from this fundamental fact:
Community governance is hard.
Or, to be more precise, building appropriate community governance
around what was proprietary software is insanely difficult.
If you look at the major Sun contributions over the years -
OpenOffice, Java, OpenSolaris, Netbeans, GlassFish, GNOME, and more recently
the purchase of MySQL, the only one of these projects which has been Sun approaching
an existing community project and participating in it is GNOME. MySQL is also
a special case, where Sun acquired GPL software.
In every other case, the projects have come from freeing a large body of
code created in a proprietary environment. And every single project I know which
was born like this has had trouble building a community. Ask
these guys.
This doesn’t just happen on its own.
When Jamie Zawinski resigned
from the Mozilla project, it was one year since the code had been freed. When
Joel Spolsky
criticised them for not shipping product, it was over two years old. When
Firefox (then Firebird)
shipped its first usable browser, Mozilla was a grand old man of 4. When Firefox
1.0 shipped, the source code had been released over 6 years beforehand.
It is much easier to get governance right when it Just Happens.
The guy who founded the project is the Boss.
A bunch of active developers fork and become
new Founding Fathers.
The company controlling
the software fully expects to pay everyone who will develop the software, and
gets outside contributors to sign away their copyright.
In all of these cases, the expectations are set by the status quo. No-one
would expect Mark Spencer to accept a feature from someone who hadn’t signed
a copyright assignment. That’s not the way Asterisk works. No-one would expect
a feature to be accepted into Linux if Linus doesn’t want it. People expect
a consensus-based approach in Inkscape.
And yet from all of what I’ve read, some people expected Sun to go from proprietary
kernel development (with a team of proprietary kernel developers, and layers
of proprietary software managing managers above them) to a bazaar overnight
(or, at the very least, very quickly). Perhaps that’s because of the way Sun
presented this to the community, perhaps it’s because certain people knew that
was an unrealistic expectation, and set Sun up to beat them over the head with
the “you’re not open” stick when they “failed” to completely open the project
in the first year.
Personally, I’d like to see as much energy going into helping Sun get things
right as is currently going into knocking every effort they make to do so on
their own. There are a great many people at Sun who don’t get it, and a great
many who do. I’d like the latter to win through.
Comments
-
pinky Says:
Great post Dave!
I also can’t understand why so many people bashing sun. Sun has
done a lot for free software and i’m really thankful for their work
on and for free software!
In the second part of your post you write a lot about a development
community and the “open source development model”. Well, i believe
that this model often creates better software but i’m not sure if
it is always the best software development model.
And even more important i don’t care about the development model.
For me it is important that the software which runs on my computer
is free software. Whether the software is developed by a small group
of people or by a large community is an side issue. The important
part is that the developer respects everybodys freedom!
Sure i would hope that they choose the development model which creates
the best “end product” but it’s a decision which the developer has
to make and i accept their decision as long as at the end we get
free software out of the development process.
- Mox Says:
So how many years does it _still_ take for the OpenOffice.org
to become a truly collaborative effort by allowing the major
non-Sun contributors to the “top” of the organisation?
Technologically this could be started by using a truly
distributed version control, like git, instead of the
high-cost-to-contribute CWS system.
-
Dave Neary Says:
Mox: Personally, I think that the time came a couple of years
ago for Sun to follow in the footsteps of the Eclipse
Foundation, set up an independent incorporation for OOo, get
501(c)6 status, and have a board of people nominated from major
contributors. OOo as a project is really too big to be easily
accessible to a volunteer community, but the project has
succeeded in gaining industry support - an initial board would
doubtless include IBM, Sun and Novell as major members, but
might also include CollabNet, the French ministry for the
interior, maybe NeoOffice and StarXpert?
In any case, the structure of a trade organisation, which
aims more to have an ecosystem than a wide-open community, seems
more appropriate for a project like OOo. It provides all of the
things which Michael Meeks has been calling for - an independant
governing body which owns trademarks and copyright, and is
answerable to companies and communities in proportion to their
contributions.
Dave.
About: MilaX is a small OpenSolaris live CD distribution. You can
run it from miniCD, bootable business card, or a USB flash drive. It is based
on Solaris Nevada. It can be installed on storage media with small capacities
like bootable business cards, USB flash drives, various memory cards, and Zip
drives. It can be installed to hard disk (UFS), or you can use a ZFS-boot installation.
Changes: This release is based on Nevada 85 and includes 99% of b85-drivers,
Gtk-Terminal, Netsurf, gFtp, Sylpheed, and gPicview. Beaver, Torsmo, and fbxkb
were added, and Dillo and aterm were removed. ZFSinstall is now included.
/sbin/sh was changed to ksh93.
Right now Sybase's enterprise value is around $2 billion, or roughly double
its current annual sales. By this measure, Sun would have to grow MySQL's revenues
to $500 million per year to bring it into sync with the purchase price. Somehow
that doesn't seem very likely, at least not in the foreseeable future.
...Sure, owning MySQL will open a few doors for the guys selling Sun boxes,
and that may lead to a few extra sales. But it's hard to see how these relationships
can translate into the large and sustained stream of new revenues Sun would
need to make the acquisition numbers work from hardware alone.
...Like most commercial open source companies, MySQL makes money by enticing
well-heeled customers to pay for an
enterprise version of its product
that comes with more bells and whistles than the
community
version it gives away for free. Both versions are available under the GPL,
but MySQL also offers a commercial license aimed mainly at OEMs and ISVs who
want to bundle MySQL with proprietary software packages. Like Red Hat, MySQL
limits access to the binaries of its Enterprise version to paying customers.
If you want a free (but unsupported) copy of MySQL Enterprise Server, you'll
have to compile it yourself.
...It appears though that the additional features of the Enterprise version
are not enough to compensate for the revenue-destroying effects of the free
Community alternative. What else could explain the surprising fact that MySQL
has quietly filled out its open source portfolio with a closed source proprietary
management software tool known as Enterprise Software Monitor? This
technically impressive product has a growing feature set that includes the
ability to monitor and manage multiple MySQL instances from a single web console.
The basic version of it comes bundled with the $1,999 per year Silver subscription
to MySQL Enterprise Server. More feature-rich versions (including replication
and memory usage management) come with the $2,999 Gold or $3,999 Platinum subscriptions.
Unfortunately for Sun Postgress now became more of a liability then asset as
it by and large overlaps (and is better) then the product for which Sun paid one
billion...
A billion dollars for a company that gives its products
away for free?
Facebook gives its products away for free, too. They
make money on ads, we make money on service, support and
infrastructure. MySQL has a big business, growing very rapidly.
Investing in the future has more value than buying the past
- which is why the latter so often comes at a discount.
What happens to your commitment to PostgreSQL?
It grows. The day before we announced the acquisition,
and within an hour of signing the deal, I put a call into
Josh Berkus, who leads our work with Postgres inside of
Sun. I wanted to be as clear as I could: this transaction
increases our investment in open source, and in open source
databases. And increases our commitment to Postgres - and
the database industry broadly. The same goes for our work
with Apache Derby, and our JavaDB.
Josh says it exactly right
on his blog - Sun wants to be the leading provider of
datacenters. Not just MySQL datacenters. Exactly.
I doubt that Oracle is a winner. when Oracle discarded Solaris in favor of linux
it made a risky move and now may need to pay for consequences. But the author is
right: the price Sun paid was very high: 20x annual revenue... Hopefully MySQL
can serve as a catalyst for hardware and software (especially Solaris) sales...
Sun CEO Jonathan Schwartz said the MySQL deal was the “most important acquisition
in the history of Sun.” But he also said the MySQL acquisition was complementary
to Sun’s JavaDB (Berkeley) and postgreSQL offerings. The latter are other open
source software [OSS] projects in which Sun is involved that compete with MySQL.
That the competing projects complement each other may be Sun’s intention but
that’s just not human nature.
... ... ...
Research 2.0 and others estimate MySQL did about $75 million in revenue in
2007. We applaud Sun’s negotiating skill. Even if MySQL only did $50 million
in 2007, the lowest estimate we have heard, it means
Sun “only” paid 20x annual revenue. In October 2007, Citrix (CTXS)
acquired Xensource for somewhere north of 100x 2007 revenue. To be fair to Citrix,
it believes it paid about 10x 2008 revenue for Xensource. But of course we won’t
know if that’s true for a year and Citrix won’t tell us if it was wrong anyway.
Sun Microsystems announced today that
it will be acquiring MySQL for $1 billion.
Sometimes the good guys get exactly what
they deserve.
At first blush, it seems an odd acquisition
for Sun. Sun, after all, is not (or was
not) in the database market. But Sun's historical
strength in the web economy, and MySQL's
current role as the heart of the web, makes
it an interesting, important step for Sun
to make.
It's nice that Solaris will the the top platform for MySQL. That's a shrewd
move. I also wonder what this means for Sun's relationship with Oracle...
announced this morning that it has agreed to acquire open source database
leader MySQL AB for $1 billion in cash and assumed
stock options. (Disclosure: I am on the board of directors of MySQL, and O'Reilly
co-produces the MySQL User Conference with
MySQL. In addition, O'Reilly produces the java.net
community site for Sun.)
This seems to me to be a great deal both for Sun and for MySQL. Anyone who
follows this blog or has heard my talks will have seen me say "Data is the Intel
Inside" of the next generation of internet applications, the very heart of Web
2.0. And of course, most of those Web 2.0 applications are built on the LAMP
stack, where M stands for MySQL, far and away the leading open source database.
Years ago, John Gage, Sun's chief scientist, made the provocative statement
"the network is the computer." And bit by bit, the industry has been realizing
that dream. What we didn't understand when we first started thinking about that
emerging network operating system was just how much it would be a data-oriented
system, such that you might more accurately say, "the network plus the database
is the computer."
The acquisition is also a great fit because Sun has staked its future on
open source, releasing its formerly proprietary crown jewels, including Solaris,
Java, and the Ultra-Sparc processor design. But even beyond those relatively
recent moves, Sun was arguably the first great open source success story, co-founded
by Bill Joy, who not only led the Berkeley Unix project but wrote the open source
TCP/IP stack on which so much of the internet was built. And even leaving out
other open source projects at the company such as openoffice.org and netbeans,
Sun has long been the single largest corporate contributor to the open source
ecosystem. (For further support for that claim, see page 51 in last year's EU
study on open source software [pdf].)
This has been a bit of a lightning courtship, and I haven't had a chance
to discuss yet with Sun CEO Jonathan Schwartz just how he plans to take advantage
of MySQL's leadership position in the open source and internet-connected database
market, but I do think that there is great potential for both companies. With
one bold stroke, Sun has reshaped both the database and open source landscape.
We're all going to be chewing on the implications for some time.
[Dec 21, 2007]
LXER interview with
John Hull - the manager of the Dell Linux engineering team
The original sales estimates for Ubuntu computers was around 1% of the
total sales, or about 20,000 systems annually. Have the expectations been met
so far? Will Dell ever release sales figures for Ubuntu systems?
The program so far is meeting expectations. Customers are certainly showing
their interest and buying systems preloaded with Ubuntu, but it certainly won't
overtake Microsoft Windows anytime soon. Dell has a policy not to release sales
numbers, so I don't expect us to make Ubuntu sales figures available publicly.
"When you take them out of the big buildings, without the imprimatur of Hewlett-Packard,
IBM and Oracle, or HP around them, they just didn't hold up."
Szulik, who took over as CEO from Bob
Young in 1999 just a few months
after its initial public offering, said
he's stepping down because of family
health issues.
"For the last nine months, I've struggled
with health issues in my family," and
that priority couldn't be balanced with
work, Szulik said in an interview. "This
job requires a 7x24, 110 percent commitment."
Szulik, who remains chairman of the
board, praised Whitehurst in a statement,
saying he's a "hands-on guy who will
be a strong cultural fit at Red Hat"
and "a talented executive who has successfully
led a global technology-focused organization
at Delta."
On a conference call, Szulik said
Whitehurst stood "head and shoulders"
above other candidates interviewed in
a recruiting process.
He was a
programmer earlier in his career and
runs four versions of Linux at home,
he said.
Moreover, Szulik said he wasn't satisfied
with more traditional tech executives
who were interviewed.
"What we encountered was in many
cases was a lack of understanding of
open-source software development and
of our model," he said. During the interview,
he added about the tech industry candidates,
"When you take them out of the big buildings,
without the imprimatur of Hewlett-Packard,
IBM and Oracle, or HP around them, they
just didn't hold up."
The surprise move was announced as
the leading Linux seller announced results
for its third quarter of fiscal 2008.
Its revenue increased 28 percent to
$135.4 million and net income went up
12 percent to $20.3 million, or 10 cents
per share. The company also raised estimates
for full-year results to revenue of
$521 million to $523 million and earnings
of about 70 cents per share.
[Dec 19, 2007] Preparing for 2008
Recession may or may not strike but it's better to be prepared:
Do some software cost cutting
- Cut the diversity of your operating systems environment.
- Use zones for consolidation.
- Eliminate all semi-useless security products. in good times enterprises
got just to many ISS sensors, firewalls, etc. You can get the same or better
level of security with free snort and you can safely replace them with free
Snort as with ISS. cutting the number of firewall can also improve security
as the total rule base became more manageable.
- Verify that enterprise monitoring systems you use worth the bucks you
pay.
Do some hardware cost cuttings:
- Unless we are talking really business critical expensive NAS solutions
like EMS do more harm then good: disks are now large and mirroring is enough
to keep data safe. also NAS introduced into data stream two cards
and a bridge all of which can be source of failures.
- Move some server to Intel 5160 CPU based servers. 5160 is probably the
most cost effective CPU on the market and you can save some money using
cheap Intel-based servers instead of more expensive Sparc-based. Excessive
zeal here can backfire as there are hidden advantages of Sparc architecture
which can became evident only after the transition.
After two years of trying to make RHEL work, Rand had to move on. He looked
closely at Solaris 10 and, after speaking with Sun engineers about a possible
migration, decided to give Sun's
Startup Essentials program a try.
"Being Linux people, we were hesitant to switch," he said. "We didn't even
consider [Microsoft]
Windows, because we are open source," said Rand. "Sun set up some virtual
servers for us to run tests, and we ported all of our apps onto those virtual
servers. We did load testing, saw that it worked well and decided to go ahead
with the migration."
Sapotek now runs Solaris 10 OS on Sun 4200 servers with 64-bit
Advanced Micro Devices Inc. Opteron quad-core processors, along with Sun's
x4500 storage unit.
The improvement is significant; with four compute nodes instead of five,
Rand has more computing power and 99.99% uptime, compared with 97% uptime with
RHEL, he said.
"With this switch, we've gone from playing in the sandbox to getting our
doctoral degree. You can't even compare Red Hat GFS to
Solaris
ZFS," Rand said. "We no longer need to do all those chores we had to do
with Linux. I can't even quantify the number of man-hours we freed by moving
to Solaris. We have so much more time to develop our software now."
This document describes how to setup
Unison
to perform synchronization between a windows laptop and a solaris system.
What I am trying to achieve is to use the Windows version of Unison, as compiled
by Max Bowsher. This version unfortunately has a problem asking for password
for the ssh account but following this document should provide an acceptable
alternative.
What I do is run Unison on the laptop and make it ssh to the solaris system
where the remote files are stored (and backed up).
For this to work, you will need to install a few
Cygwin packages (for
ssh) and manually install Unison for windows and at last, set it up so we can
avoid the bug mentioned above.
Subject: Solaris and Dell ( Nov 16, 2007, 08:03:06 )
While I would agree that the majority of commodity servers that are non
MS based, have moved to Linux, I do see Open Solaris becoming a real option
for those shops that have relied on Solaris application stacks for years.
Those who are entrenched with the Solaris experience, are less likely to move
away from it to adopt Linux, if they can have the same old stable OS on commodity
hardware that they have been using for so many years. Why do I say this? Experience.
As a Software Architect for a professional services provider and as one who
has worked almost exclusively within the Enterprise IT world (where the corporation
size is greater than 20,000 employees), I have seen many of these shops who
may have migrated to Linux servers, not do so, since they have been Solaris
customers for years and they see no reason to move away from an OS that has
served them extremely well in the past. Solaris 10 and/or Open Solaris basically
offer these companies the means to continue to use what they know and are comfortable
with and at the same time, maintain the support relationships with Sun they
have enjoyed through the years.
Solaris 10 and Open Solaris on x86-64 makes it easy
to enjoy the price advantages of Linux without the need to experiment or change.
To me, this is the advantage of the Sun-Dell deal that is provided for Sun customers.
Many of these same Enterprises that have a relationship with Sun, also have
a relationship with Dell. What this provides these customers is a means to price
shop on purchases between Sun and Dell, without feeling that they are gambling
with support issues. In other words, it provides the Solaris
customer the advantage of lowering TCO, without losing the skills and support
they have enjoyed through the years.
In fact, I will have to say, that the largest threat to Linux completely taking
over the server farm, is Open Solaris. MS will keep a chunk of this share via
Virtualization, but I cannot see IT shops continuing to invest in the mantra
of Microsoft's, "write here run only here", philosophy for too much longer.
Once IT shops discontinue employing a mono-platform development philosophy,
MS will have a much more difficult time holding onto its server side market
share. I can see a open source Solaris however, bantering for the mindset of
those looking at an open source OS strategy in the future. Of course how successful
Sun is in keeping Solaris as a real competitor to Linux is how well they develop
a real relationship with the open source development committee and how innovative
they are with Open Solaris. This is the larger challenge Solaris now has. If
they cannot keep Solaris meeting or exceeding the feature sets of Linux, the
battle is over. If however, they can best Linux in Enterprise level features
(they have a slight advantage here now), then Open Solaris and Linux will become
the dual heavyweight contenders.
Personally, I have no preference. I am as content with the one as I am with
the other. However, Linux has a huge advantage over Solaris for Desktop share
and that could tip the scales in Linux's favor for the server room as well.
|
Sun Microsystems and Dell announced Wednesday November 14, 2007 a distribution
agreement under which Dell will distribute Sun's Solaris 10 operating system on
all Dell servers and first of all PowerEdge servers. New deal marks the first time
Solaris will be supported by Dell
Dell founder and CEO Michael Dell and Sun Microsystems CEO Jonathan Schwartz
made the announcement during a joint appearance at the Oracle OpenWorld 2007
conference.
The agreement means that customers buying a Dell servers get the option of
installing Solaris or OpenSolaris. Customers picking one of these operating
systems will get support from Sun's online support organization through Dell,
making the experience seamless for the customer.
This marks the first time that Sun's home-grown, Unix-based operating systems
will be sanctioned for use in any kind of Dell hardware. The two companies have
been rivals in the server business for more than 12 years.
The new agreement means that Dell will test, certify, and optimize Solaris
and OpenSolaris on its rack and blade servers and offer them as one of several
choices in the overall Dell software menu.
Dell already supports Windows as well as both Red Hat and SUSE Linux in all
its rack and blade servers.
According to terms of the agreement, customers will be able to freely download
OpenSolaris from the Dell website. Sun has used proprietary Solaris since the
1980s as its chief operating system for workstations and servers; it released
the freely available open source version, OpenSolaris, in June 2005.
The new partnership opens two new markets for both companies: Dell now can
sell its hardware into both the proprietary Solaris development world and the
growing open source OpenSolaris community. Sun will get its software into numerous
new systems and obtain a new gateway into the SMB (small and medium-size business)
market through Dell's brand.
At this time, few SMBs use Sun hardware, since the company has focused almost
exclusively in the past on the large enterprise market. This will start new
conversations as Sun starts coming out with more mid-tier hardware.
"There are three main reasons we are doing this," Rick Becker, vice president
of solutions in the Dell Product Group, told eWEEK. "No. 1 is Sun's new and
strong commitment to x86 systems; secondly, a lot of people are already using
the Solaris operating systems; and three, our existing customers are asking
for this option."
The deal gives corporate developers the option of using Sun's bread-and-butter,
Unix-based enterprise operating system -- which includes the fast ZFS (Zettabyte
File System) -- in Dell boxes, which are generally less expensive than most
other servers and used in hundreds of thousands of enterprise and SMB systems
worldwide.
Becker told eWEEK that customers choosing Sun operating systems will get
support via Dell in a seamless manner.
Dell appears to be getting the better part of the deal, at least at the outset.
Dell will get the margins from selling the hardware, but ostensibly, Sun looks
like it will be getting only service contracts from those who choose to use
either of the Solaris options.
One of the first customers for this will be the U.S. Navy, Becker said.
"sysstat" complements Solaris' system tools for performance analysis. It
presents all key performance metrics on a VT100 terminal and has the possibility
to toggle its view between different hosts.
The ability to run linux applications and assign CPUs to zones are provided
in Solaris 10 8/07
Solaris Container Manager provides additional zone management features that
are implemented in Solaris 10 8/07.
Managing Branded Zones
...The branded zone (BrandZ) framework enables you to create non-global zones
that contain non-native operating environments used for running applications.
Currently linux is supported. All brand management is performed through extensions
to the current zones structure.
For more information about branded zones, see
System Administration Guide: Solaris
Containers-Resource Management and Solaris Zones.
To Create a Branded Zone
- Navigate to the New Zone Wizard.
The New Zone wizard appears.
- Work through the wizard to reach the step "Provide zone creation attributes".
- Choose lx from the Zone Brand drop-down list.
The lx value for zone brand is available only on Solaris 10 8/07 x64
systems.
The zone brand determines the scripts that are executed when a zone is
installed and booted and identifies the correct application type at application
launch time. The possible values of zone brand are:
- Native - Specifies that the zone contains the same operating environment
as the parent host.
- lx - Specifies that the zone contains a Linux environment.
- Type the image path and install arguments and click Next.
- Specify the system configuration file.
This file is required to provide the attributes that are required for
zone management. You need to create this file and specify the location here.
Assigning Dedicated CPUs to a Zone
You can assign dedicated CPUs directly to a zone. When the zone requests
a specific number or range of CPUs, the system creates a temporary resource
pool with the name SUNWtmp_zonename. The temporary
resource pool assigns these CPUs to the zone. When the zone shuts down, the
resource pool releases these CPUs.
To Assign Dedicated CPUs to a Zone
You can assign dedicated CPUs to a zone only on Solaris 10 8/07.
- Navigate to the New Zone Wizard.
The New Zone wizard appears.
- Work through the wizard to reach the step "Select a Resource Pool".
- Select the Enabled check box for dedicated CPU allocation.
- Type the number or range of CPUs in the Number of CPU or Range field.
For example, type 3 or 1-5.
From:
|
Glynn Foster <Glynn.Foster-UdXhSnd/wVw-AT-public.gmane.org> |
|
To: |
|
Open Solaris <opensolaris-discuss-xZgeD5Kw2fzokhkdeNNY6A-AT-public.gmane.org>,
OpenSolaris Announce <opensolaris-announce-xZgeD5Kw2fzokhkdeNNY6A-AT-public.gmane.org>,
Indiana Discuss <indiana-discuss-xZgeD5Kw2fzokhkdeNNY6A-AT-public.gmane.org>,
advocacy-discuss-AT-op |
|
Subject: |
|
[indiana-discuss] Project Indiana milestone reached! |
|
Date: |
|
Thu, 01 Nov 2007 16:32:34 +1300 |
I'm very pleased to announce that the first milestone of Project Indiana is now
available - called OpenSolaris Developer Preview.
It's available for download at
http://dlc.sun.com/osol/indiana/downloads/current/in-prev...
This is an x86-based LiveCD install image, containing some new and emerging
OpenSolaris technologies. This may result in instabilities that lead to system
panics or data corruption.
Among the features contained in this release are
o Single CD download, with LiveCD 'try before you install' capabilities
o Caiman installer, with significantly improved installation experience
o ZFS as the default filesystem
o Image packaging system, with capabilities to pull packages from
network repositories
o GNU utilities in the default $PATH
o bash as the default shell
o GNOME 2.20 desktop environment
For more details about the system requirements along with some basic user
documentation, see -
http://opensolaris.org/os/project/indiana/resources/getit/
and the release notes
http://opensolaris.org/os/project/indiana/resources/rn/
This milestone preview shows the results of many months of engineering work
through the collaboration of several projects on opensolaris.org. I would like
to thank to those people who have been involved, and offer my congratulations
for reaching this successful milestone.
Report Bugs
===========
We are very interested in hearing feedback about your experiences with this
release. In particular, if you have issues installing on your hardware we would
love to know.
If you would like to provide feedback, see our bug reporting page for details on
how to do that -
http://www.opensolaris.org/os/project/indiana/resources/r...
About Project Indiana
=====================
Project Indiana is working towards creating a binary distribution of an
operating system built out of the OpenSolaris source code. The distribution is a
point of integration for several current projects on OpenSolaris.org, including
those to make the installation experience easier, to modernize the look and feel
of OpenSolaris on the desktop, and to introduce a network-based package
management system into Solaris.
http://www.opensolaris.org/os/project/indiana/
Rock on!
Glynn
On behalf of Project Indiana Team
© 1996 University Technology Services, The Ohio State University, 1971 Neil
Avenue, Columbus, OH 43210.
All rights reserved. Redistribution and use, with or without modification,
are permitted provided that the following conditions are met:
- Redistributions must retain the above copyright notice, this list of
conditions, and the following disclaimer.
- Neither the name of the University nor the names of its contributors
may be used to endorse or promote products or services derived from this
document without specific prior written permission.
THIS PUBLICATION IS PROVIDED "AS IS" WITHOUT WARRANTY OF ANY KIND. THIS PUBLICATION
MAY INCLUDE TECHNICAL INACCURACIES OR TYPOGRAPHICAL ERRORS.
This publication is available via the Internet as:
http://wks.uts.ohio-state.edu/sysadm_course/sysadm.html.
Also available via the Internet is Introduction to Unix:
http://wks.uts.ohio-state.edu/unix_course/.
Acknowledgements: The author wishes to thank the following for helpful advice
and discussions related to the material presented in this document: Harpal Chohan,
Bob DeBula, Bob Manson, Steve Romig, and Bill Yang.
Table of Contents
-
PART I - Introduction
-
___CHAPTER 1 - Overview
-
___CHAPTER 2 - Disk Structure and Partitions
-
___CHAPTER 3 - Devices
-
___CHAPTER 4 - The UNIX File System
-
___CHAPTER 5 - File System Management
-
___CHAPTER 6 - Startup and Shutdown
-
___CHAPTER 7 - Operating System Installation
-
___CHAPTER 8 - Kernel Configuration
-
___CHAPTER 9 - Adding Hardware
-
___CHAPTER 10 - Special Files
-
___CHAPTER 11 - System Directories
-
___CHAPTER 12 - User accounts
-
___CHAPTER 13 - Daily System Administration
-
___CHAPTER 14 - Administration Tool & Solstice Adminsuite
-
___CHAPTER 15 - Package Administration
-
___CHAPTER 16 - Backup Procedures
-
PART II - Network Services
-
___CHAPTER 17 - Service Access Facility
-
___CHAPTER 18 - The Network
-
___CHAPTER 19 - Network Administration
-
___CHAPTER 20 - Distributed File System Administration
-
___CHAPTER 21 - Network Information Services (NIS and NIS+)
-
___CHAPTER 22 - Adding Clients
-
PART III - Selected Topics
-
___CHAPTER 23 - Usenet
-
___CHAPTER 24 - Useful Utilities
-
___CHAPTER 25 - Print Service
-
___CHAPTER 26 - Mail
-
___CHAPTER 27 - World Wide Web
-
___CHAPTER 28 - System Security
-
___CHAPTER 29 - Secure Shell, SSH
-
PART IV - Summary
-
___CHAPTER 30 - Summary of SunOS/Solaris Differences
-
___CHAPTER 31 - UTS UNIX Workstation Support
1.What exactly does this cover? Which network protocols? Which data
formats?
See the
EC ruling [europa.eu] (PDF), especially article 999 on page 277:
(999) Microsoft should be ordered to disclose complete and accurate
specifications for the protocols used by Windows work group servers
in order to provide file, print and group and user administration services
to Windows work group networks. This includes both direct interconnection
and interaction between a Windows work group server and a Windows client
PC, as well as interconnection and interaction between a Windows work
group server and a Windows client PC that is indirect and passes through
another Windows work group server. The use of the term specifications
makes clear that Microsoft should not be required to disclose its own
implementation of these specifications, that is to say, its own source
code. The term protocol relates to the rules of interconnection and
interaction between instances of the Windows client PC operating system
and the Windows work group server operating system.
Also interesting:
(1008) The requirement for the terms imposed by Microsoft to be reasonable
and non- discriminatory applies in particular: [...] there is a need
to ensure that potential beneficiaries will have the opportunity to
review, themselves or through third parties designated by them, the
specifications to be disclosed; Microsoft should be able to impose reasonable
and non-discriminatory conditions to ensure that this access to the
disclosed specifications is granted for evaluation purposes only;
[...] to any remuneration that Microsoft might charge for supply; such
a remuneration should not reflect the strategic value stemming from
Microsoft s market power in the client PC operating system market or
in the work group server operating system market;
The decision does not seem to give a hard number for how much MS may charge
for disclosure of the specs.
Just venting here...
What you describe is utter stagnation - "Microsoft can not change the protocol
without pissing off many companies
..."
This is an absolute formula for zero growth. It is one of the things that causes
a lot of Linux development to be done in fits and starts where something stagnates
for a long period of time because "Oh, we can't manage change in THIS area."
This is a formula for a repeat of where we are with SMTP today.
All change is not bad. There are ways of implementing changes in established
protocols if the original protocol allows for it and it is done carefully.
Today, we are stuck with SMTP and no replacement is anywhere on the horizon.
Why? Because the nobody wants to manage the change.
Microsoft is not the enemy. They can be a partner and must be if there is to
be any real progress by anyone except Microsoft. Calling them the enemy, refusing
to work with closed-source software and just trying to be obstinate will result
in Microsoft being the only choice far, far into the future.
What if tomorrow KFC was forced to give up the eleven herbs and spices
used in their secret recipe? Don't laugh, it may happen someday. I'm
no fan of Microsoft, but can't people just create something better?
Sure, there are anti-trust laws, but whatever happened to beating someone
in the market by creating a better product? Years back, Netscape tried
to sue Microsoft because they felt their browser was unfairly marketed
since it came with the system. Today, Mozilla is proving Netscape dead
wrong and is almost 20% of all browsers while Microsoft is down to almost
60%. It's a sad day when the lawyers are writing better code than the
developers.
First, open source software developers will be able to access
and use the interoperability information. Microsoft will not assert
patents against non-commercial open source software development
projects.
The opposite of "open source" is not "non-commercial". There are
commercial open-source prjects, and non-commercial closed-source
projects. It is absolutely vital that these interfaces be as unencumbered
as genuinely open-systems protocols are.
Second, the royalties payable for this information will be reduced
to a nominal one-off payment of 10,000 euros.
US$14,000 is not "nominal".
Third, the royalties for a worldwide license including patents
will be reduced from 5.95 percent to 0.4 percent, far less than
the 7 percent originally demanded by Microsoft.
Getting a European court to acknowledge the validity of their software
patents at all is a major win for Microsoft.
And the way they did this means that there's not a hope of an avenue
to try and appeal this appalling result. Microsoft has completely
won this round in their ongoing battle against open systems and
open source.
If this is Microsoft "bowing", they're facing away from the bench
when they do it, and mooning the EU.
Microsoft agreed to license proprietary information on how Windows
shares files and printers to end three years of legal wrangling over
a 2004 antitrust order. The accord will help Red Hat Inc., the world's biggest
seller of Linux systems, and Sun Microsystems Inc. offer replacements for Windows.
``These changes in Microsoft's practices will profoundly affect software
industries,'' European Competition Commissioner Neelie Kroes told reporters
in Brussels today. ``I sincerely hope that we can just close this dark chapter
of our relationship.''
The accord furthers Microsoft's bid to resolve legal disputes worldwide that
have been weighing on its shares. The company last week dropped its appeal of
an antitrust decision in South Korea and today said it won't challenge a court
decision last month upholding the EU decision. It's also seeking to end five
years of U.S. court supervision for illegally protecting its near-monopoly on
PC software.
Microsoft rose 34 cents, or 1.1 percent, to $30.51 at 4 p.m. New York time
in Nasdaq Stock Market trading. The stock has gained 26 percent since the EU
imposed a record 497 million-euro ($703 million) fine and ordered the company
to change its business practices in March 2004. The Nasdaq Composite Index has
gained 44 percent in the same period.
Trade Secrets
Under the 2004 decision, Microsoft had to disclose information to rivals
and sell a version of Windows without a built-in video and audio player. The
company resisted licensing data to open-source developers, who give away the
software's source code, or the underlying instructions, because it would violate
trade secrets and patents.
Kroes said open source products are ``virtually the only alternative'' to
Microsoft, which has more than 70 percent market share for workgroup server
software.
Microsoft got $4.5 billion in sales from its Windows Server software in its
most recent fiscal year. Since 2002, the product's sales have grown at an annual
rate of 13 percent, on average.
That might simplify achieving Solaris interoperability with active directory.
Microsoft finally admitted defeat
in its nine-year battle with the European
Commission on Monday, agreeing to allow
competitors access to technology that Brussels
said would create more innovation in the
software market.
The US software developer agreed to comply
with the EU antitrust regulator’s finding
that it was abusing its dominance, upheld
by the European Court in 2004. The result
would be lower prices and more choice for
customers, the Commission said.
“I welcome the fact that
Microsoft has finally undertaken
concrete steps to ensure full compliance
with the 2004 decision,” Neelie Kroes, competition
commissioner, said in Brussels. “It is regrettable
that Microsoft has only complied after a
considerable delay, two court decisions
and the imposition of daily penalty payments.”
Steve Ballmer, Microsoft chief executive,
agreed early on Monday to make it easier
and cheaper for rivals to link their products
to some classes of its software. While only
affecting software for so-called “workgroup”
servers, widely used but low-value software
that manages jobs such as printing from
networked computers in an office, the decision
is the first tangible result of Microsoft’s
defeat before the European Court of First
Instance, Europe’s highest court, last month.
WebSphere Application Server 6 on Sun Solaris 10, SA-7777-R01
Open
Starts 05 Nov 2007, ends 07 Dec 2007 (5 weeks) and requires 5 residents.
Contains several interesting tables, for example support costs comparison table:
Jonathan Schwartz fundamental question raise another question: to what extent
openness is a marketing advantage (in reality using or even stealing source code
from the huge software project is very difficult as it is more like a huge infrastructure
or even organism of which source code is just a small part). At the same time open
source is an insurance for the user even if then never use it.
There are four fundamental questions/topics in open source:
- Open-source licenses and the availability of source code;
- The impact of free (as in cost) software;
- The value of brand. As Red Hat knows, Red Hat is indomitable because
of its brand, not its source tree;
- Who's asking? The answer you give to an 8-year-old is different from
the one you'd give to a CIO. This last topic provides the answer to the
open-source revenue question.
Why? Think about this: In a year where Sun arguably moved more aggressively
to give away more free software than any other company, we grew our software
business by 13 percent. It was the fastest-growing business at Sun (and
doesn't even include Solaris, which we don't yet break out). We pumped out more
software last year than we have in the history of the company. We gave it away.
And yet our software business grew by 13 percent.
Submitted by
Jeremy on August 7, 2007 - 9:26am.
In a recent lkml thread, Linus Torvalds was involved in
a discussion about mounting filesystems with the
noatime
option for better performance, "
'noatime,data=writeback'
will quite likely be *quite* noticeable (with different
effects for different loads), but almost nobody actually
runs that way." He noted that he set O_NOATIME when
writing git, "
and it was an absolutely huge time-saver
for the case of not having 'noatime' in the mount options.
Certainly more than your estimated 10% under some loads."
The discussion then looked at using the
relatime
mount option to improve the situation, "
relative atime
only updates the atime if the previous atime is older than
the mtime or ctime. Like noatime, but useful for applications
like mutt that need to know when a file has been read since
it was last modified." Ingo Molnar stressed the significance
of fixing this performance issue, "
I cannot over-emphasize
how much of a deal it is in practice. Atime updates are
by far the biggest IO performance deficiency that Linux
has today. Getting rid of atime updates would give us more
everyday Linux performance than all the pagecache speedups
of the past 10 years, _combined_." He submitted some
patches to improve
relatime, and noted about
atime:
"It's also perhaps the most stupid Unix design idea
of all times. Unix is really nice and well done, but
think about this a bit: 'For every file that is read
from the disk, lets do a ... write to the disk! And,
for every file that is already cached and which we read
from the cache ... do a write to the disk!'"
Not a bad start. Recently Scarc was complety off SPECint and SPECfp charts.
Sun Ultra SPARC T2 @1.4GHz got something like:
- SPECint_rate2006 78.3
- SPECfp_rate2006 62.3
It is still unclear when it will ship, whether all T2 chips will be 1.4GHz and
how much will it cost.
- The UltraSPARC T2 processor is the fastest commodity processor with two
world-record single-chip SPEC CPU results, based on tests that delivered 78.3
est. SPECint_rate2006 and 62.3 est. SPECfp_rate2006
- Powered by fewer than 95 watts (nominal) with less than 2 watts per thread,
this next generation UltraSPARC T2 processor boasts the most functionality,
and lowest wattage per core and thread of any processor in its class
- Integrated multithreaded 10 Gb Ethernet networking
- Integrated PCI Express I/O expansion
- Integrated floating point and cryptographic processing units per core
July 2007 |
BigAdmin
This article provides information on performance tests that compared the
Sun Fire V480 server to the Sun Fire T2000 server in an environment that runs
an Oracle 10g Real Application Clusters (RAC) database to store and serve the
data required to run a corporate web site.
Contents
This article covers the following topics:
Solaris x86-64 platform: AMD64, EM64T, and compatible processors - any hardware
that is explicitly compatible and fully capable of running the specified operating
system, all the corresponding supporting software shown below, and any associated
applications unmodified.
64 bit
- Sun Java 2 Platform Standard Edition, Version 5.0
- IBM 64-bit SDK for Solaris, Java 2 Technology Edition V5.0 (SR1 or above)
[Only if the JDK is supplied with another IBM product]
Notes:
- JDBC/XA usage is not supported on this platform.
- Sun platform with 64-bit hardware, 32-bit and 64-bit WebSphere MQ APIs
are supported and the OS must have a 64-bit kernel.
- On platforms where WebSphere MQ provides a 64-bit product, both 32-bit
and 64-bit MQ APIs are supported and the OS must be capable of running 64-bit
applications.
- On 64-bit platforms where WebSphere MQ provides both a 32-bit and a
64-bit product, there is NO migration path from V6 32-bit Queue Managers
to V6 64-bit Queue Managers. You are strongly advised to install WebSphere
MQ 64-bit products on 64-bit platforms as there is a migration path from
V5 32-bit products to V6 64-bit products.
- The 'MQ Explorer' is not supported on this platform.
-
Solaris Zones support information.
- For details of the Support Statement for JCA (on WebSphere MQ V6.0.2.1
or later only) please see
here.
IBM Software on UltraSPARC T1 License Policy
IBM announced their
new Software license policy for UltraSPARC T1 chip based Sun Servers.
A 4 core or 6 core T1 chip requires 2 software license entitlements
and a full 8-core T1 chip requires 3 software license entitlements.
(Its a win-win considering before this announcement it was 8 software
license entitlements for a 8-core T1 chip.
One thing to also note that a Sun Fire X4200 /X4100 with dual core
Opterons require 1 license entitlement for every dual core chip. (Read
the foot note for descriptions of "Multiple server families" which includes
SunFire x64 servers.
The best way to distribute the news is to point people to the
PDF Copy of the IBM Software License Policy
Time to run WebSphere on UltraSPARC T1 chip based Sun Fire T1000
or Sun Fire T2000
|
Perhaps the easiest way to communicate
the underlying concepts of ZFS is a comparison the Sun developers drew during
the design stages of the file system back in 2001. When you add RAM to a server,
you don’t partition it and allocate one DIMM to this application and another
DIMM to that application; you throw all of the RAM into a pile and let the memory
manager decide who gets what and when. That simple, pragmatic view forms the
basis of ZFS: There are no partitions and no fixed block sizes, no file system
consistency check, no RAID initialization procedure, and no inodes – there’s
just a pile of disk with ZFS in between.
... ... ...
There’s far more to ZFS than
is possible to cover in this space, so I’m hitting the high points. Starting
with the essentials, ZFS is comprised of three parts. The ZPL (ZFS POSIX Layer)
runs at a high level, taking instruction from the OS on I/O requests. Below
that is the DMU (Data Management Unit) that takes those instructions and translates
them into transaction batches. Rather than requesting data blocks and sending
single write requests, ZFS batches these into object-based transactions that
can be optimized before any disk activity occurs. Once this is done, the batches
are handed off to the SPA (Storage Pool Allocator) to schedule and aggregate
the raw I/O. The copy-on-write basis of I/O transactions, coupled with checksums
performed on a per block basis, precludes the need for journaling. An abrupt
power loss will be recoverable at any point.
Perhaps another good example
would be to illustrate how ZFS handles simple disk mirrors. In a traditional
two-disk mirror, reads from the mirror are handled in a round-robin fashion
to increase read times. This means that if there’s bit rot on one disk but not
on the other, there's a fifty-fifty chance that data requested by an application
will be invalid. With traditional RAID configurations, this data corruption
will be largely unnoticed by the underlying layers, but the application will
certainly realize that there’s a problem. ZFS overcomes data corruption by checksumming
each block as it’s returned from disk. If there’s a disparity between the 256-bit
checksum and the block, ZFS will terminate the request and pull the block from
the other member of the mirror set, matching the checksums and delivering the
valid data to the application. In a subsequent operation, the bad block seen
on the first disk is replaced with the valid data from the second, essentially
providing a continuous file system check.
But aren’t checksums expensive?
Yes. Well, at least they used to be. In the era of multicore CPUs, delegating
a single core of a CPU to performing checksums still leaves plenty of horsepower
to handle everything else. The benefits offered by this form of I/O consistency
validation eclipse the performance hits on modern hardware, and judging by my
performance tests, it’s certainly not an issue.
... ... ...
ZFS has a number of neat tricks
for managing numerous drives. Because all disk is thrown into a single pool,
adding drives to existing arrays is instantaneous, and it requires no re-initialization.
During quiescent periods, ZFS will reallocate the data across all disks for
better performance, even while making newly added storage immediately available,
with writes crossing all drives and reads coming from the original array members.
It appears that Sun also gave
careful consideration to disk workload profiling. Server file systems are commonly
asked to handle multiple sequential requests to single files. At first blush,
these calls may appear to be random I/O, but a closer look will often reveal
they are not so random. ZFS can smooth this type of workload with intelligent
read-ahead caching at the block level, resulting in significant performance
gains for streaming media and for some database workloads.
Another facet of the advanced
I/O scheduling in ZFS is request prioritization. When a system is I/O bound,
it’s generally due to the disk not keeping up with requests, or major swap operations.
Once those requests stack up, basic system interaction slows to a crawl, and
there’s nothing more frustrating than trying to kill the misbehaving process
with a command that takes forever to run because it needs to be fetched from
the very same disk that the runaway process is thrashing. Because ZFS gives
reads priority over writes, the read necessary to execute the kill command in
these cases gets pushed to the front of the queue, allowing order to be restored
in a timely manner.
Smooth
snapshots, security
As you would expect, ZFS incorporates
snapshots with simple one-line CLI commands, and it allows snapshots to be addressed
in both read-only and read-write forms. Rollbacks and individual file inspection
in snapshots are also easy to do. Further, ZFS has integrated rsync-like file
synchronization, allowing for truly different backup methods, such as piping
raw file system data across SSH connections to backup servers with enough smarts
to be usable across high-latency links.
There’s also the not-so-small
matter of ACLs, which ZFS handles with standard POSIX-compliancy and allow/deny
inheritance. Checksumming is a boon from a security standpoint as well: Because
every block has a checksum, data can’t be modified at that level without detection.
Oh, and did I mention that ZFS can also sit on top of other storage elements,
such as iSCSI LUNs (logical unit numbers) and swap volumes? Sun says its engineers
have subjected ZFS to more than a million forced, violent crashes in the company's
labs without losing data integrity or leaking a single block. I haven’t witnessed
such a crash, but I have to say I believe Sun’s claims.
Indiana will fit on a single CD and be updated
every six months, Foster said. "With a focus
on the user experience, it is hoped that with
wide distribution, the OpenSolaris ecosystem
will grow, providing valuable feedback to the
project."
And although Foster said the project is intended
to be grassroots and consensus-driven, "there
may be a real need for a sole arbiter,
Ian Murdock," who is Sun's chief operating systems
officer and a founder of the Debian version
of Linux.
It does not make sense to spend significant part of your life installing open
source applications ;-)
Solaris Express Developer Edition is an OpenSolaris-based
distribution for x86 that includes the latest tools, technologies, and platforms
to create applications for Solaris OS, Java Application Platform, and Web 2.0.
Available at no-cost, Solaris Express Developer Edition is regularly updated
to incorporate new functionality to help application developers create better
applications -- faster. Developers can create high performance applications
using this distribution and
deploy to Solaris 10 OS.
Develop your applications using Solaris Express Developer Edition and
deploy to Solaris 10. For applications that use Solaris APIs, we encourage
you to download and use the Solaris Ready Test Suit to verify use of
Solaris 10 APIs. In addition, you should do your final build on a Solaris
10 server before deploying.
Developer support options are available for code support, programming and
technical assistance. Recognized industry-wide, Sun offers
developer training
and certification courses for Solaris, Java, and
Web 2.0 developers.
The 2/07 release of Solaris Express Developer Edition is only for x86-based
laptops and desktops. Developers on SPARC systems can obtain similar functionality
by downloading the latest Solaris Express Community Edition build 55 (CD)
or (DVD)
and installing
Sun Studio 11 for OpenSolaris and
NetBeans
IDE 5.5 with
NetBeans Enterprise Pack 5.5. Future Solaris Express Developer Edition releases
will include support for both x86 and SPARC platforms.
VMware for Solaris Express Developer Edition is also available.
-
Sun Studio 11 for OpenSolaris - the latest enhancements and patches
for Solaris Express distributions. Record-setting C, C++, and Fortran compilers,
IDE, and integrated tools
-
NetBeans IDE 5.5 with Enterprise Pack 5.5 - an open-source IDE for desktop
and enterprise Java development including the
GlassFish V1 application
server.
- Solaris + AMP (SAMP) -
Learn how to set up an optimized and observable stack using Solaris Express
Developer Edition to develop your AMP applications. Note: For those who
downloaded the AMP offering, Cool Stack v1.0.2, you need to follow these
post-installation
steps on Solaris Developer Express Edition. For those that have downloaded
Cool Stack v1.1, no additional post-installation steps are necessary.
-
Solaris OS, Desktop and Applications - Based on Solaris Express Community
Edition build 55, this release includes many desktop and operating system
enhancements. The highlights include new tools like Firefox, Thunderbird,
and StarOffice 8.
- Open Source Applications- over 150 open source applications including
MySQL, Python, and PostgreSQL
The key idea of SMF is to provide XML definition file that serves as an envelope
for actual scripts that invoke the services and which documents some properties
and context in which each typical operation (start, stop, restart) should be performed
and path tot he script itself which now can be part of the application. That gives
the possibility to execute those operations more intelligemently, for example introduce
the concept of dependencies.
But the problem is that XML template used smells with overcomplexity and slightly
lessens the transparency which was the main advantage of rc* files approach...
Availability of svcs -a functionality cannot be legitimately
sited as an advantage as it can be easily implemented with th4 old approach (as
was done in Red Hat with services command).
I am not sure that conversion of init scripts to Perl cannot achieve the same functionality
at lower cost.
The Service Management Facility is a new, unified model for services and
service management that is included in the Solaris Operating System. SMF provides
a deeper, more functional view into the processes managed during startup and
shutdown of a Solaris system. In addition, processes managed through SMF can
have dependencies and they are monitored to allow for restarts if a process
fails or is improperly stopped.
SMF is a core part of the predictive self-healing technology available in
the Solaris 10 OS, and it provides automatic recovery from software and hardware
failures as well as administrative errors. In addition, SMF-managed services
can be delegated to non-root users. Finally, SMF is a follow-on to the legacy
method of starting and stopping services, though /etc/rc
scripts will continue to run when present for backward compatibility.
Deployment of services through SMF provides a much more consistent and robust
environment. First, users can query the Solaris OS with a simple command (svcs
-a) to determine if a service is running, instead of attempting a connection
and wondering if the connection will succeed. Additionally, critical services
can be restarted automatically in the event of a problem, such as someone inadvertently
killing a service, a bug causing a core dump, or other process failures occurring.
Further, SMF provides detailed and common logging as well as robust error handling
to prevent services from hanging after a system state change. Please see the
man page
for smf(5) for more information.
After a typical software installation, there can be a half dozen or more
processes that need to be started and stopped during system startup and shutdown.
In addition, these processes may depend on each other and may need to be monitored
and restarted if they fail. For each process, these are the logical steps that
need to be done to incorporate these as services in SMF:
- Create a service manifest file.
- Create a methods script file to define the start, stop, and restart
methods for the service.
- Validate and import the service manifest using
svccfg(1M).
- Enable or start the service using
svcadm(1M).
- Verify the service is running using
svcs(1).
Using SAS processes as an example, we will create two services, one for the
SAS Metadata Server (OMR) and one for the SAS Object Spawner. In this example,
the Object Spawner cannot attempt to start before the OMR is started and should
be stopped before the OMR is stopped.
It looks like JBoss performance can be significantly improved by proper tuning
of TCP/IP stack and system parameters.
Contents
In a recent SearchOpenSource.com article on
synching Linux servers with AD, users told us that using Sun NIS was a "big
no-no" in terms of Sarbanes-Oxley compliance. What have you heard on this?
Sigle: Many of those users are today moving from [Network Information
Service] to LDAP. This is because with LDAP you get native security built into
it, like SSL. With a customer I visited just last week, a large telecom, they
had 20 different NIS domains, and they were planning on consolidating those
into one infrastructure. They were putting those all into LDAP, centralized
LDAP. Their domains will all still have multiple domain names but will instead
be centralized into an LDAP tree.
Preview what you're going to be doing at LinuxWorld and why IT managers
might want to attend.
Sigle: Basically it will be [about] identity management and access
management options in the open source space conducted in a panel format with
Gianluca Brigandi, Founder and System Architect of the JOSSO Project, and Anthony
Nadalin from IBM. We will cover enterprise-to-customer relationships, business-to-business
relationships, and so on in the ID space. Eventually, the conversation will
end with a slide that I call the alphabet soup. It will list all the current
identity standards like OpenSSO, JOSSO, OpenID -- all the buzzwords.
Could you provide a little perspective on some of these standards, like
Sun's OpenSSO for example?
Sigle: Sun has been an industry lead with Directory Server, all the
way back from the Netscape days. Many enterprises and many telecoms in the market
run Directory Server for their customers, and many large telecoms run millions
of identities in Directory Server. About a year ago, Sun architected a new Directory
Server with all those aforementioned standards in mind. Sun then donated the
code to the open source community. At some point in time, the plan is for Sun
to take a snapshot of the open code, wrap support around it, and that will most
likely be the next version of a directory server we support as company. That's
one to two years away however. For now with OpenSSO, we took [Directory Server]
and its access management capabilities and basically released all the source
code. Going forward it will be the same scenario as Directory Server: we'll
release a commercial snapshot of OpenSSO in the future.
Has there been any user confusion regarding the number of standards?
Sigle: In the past I worked with telecom customers and I heard that
complaint all the time. Customers wanted to know how all these standards were
going to talk to one another. Even if we delve into one of the collaborative
efforts like the Liberty Alliance [which is comprised of Sun Java System Access
Manager, OpenSSO, Lasso (Liberty single sign on) and HP Select Federation],
there are different phases and specs. There's a bunch of stuff in there, and
a lot of these standards drive toward the same goal. Recently we have start
Re:Place for GNU?
(Score:2)(http://www.infinadyne.com/)