Solaris

Ben Rockwood wrote a great article about the usage of Crossbows vnics and virtual switches in conjunctions with zones: Crossbow Experiements and Elation. A must-read.

Ars Technica writes in "A look back at the open source victories of 2008":

Sun released the very first version of OpenSolaris this year. OpenSolaris, which emerged from Project Indiana, was created to build a user-friendly desktop distribution on top of the open source Solaris platform. Despite some early friction, the project shows a lot of promise. The second release, which occurred earlier this month, included some impressive functionality, such as a new ZFS snapshot visualization feature.

Nice to see OpenSolaris in such an article

Das neue Jahr hat begonnen, Zeit also die HHOSUG zu startenn. Obwohl die Usergruppe Hamburg im Namen traegt, ist das doch eher auf die Metropolregion Hamburg bezogen. Um erstmal einen gemeinsamen Kommunkationskanal aufzubauen, habe ich bei XING um die Einrichtung einer XING Gruppe gebeten. Gestern ist diese dan auch eingerichtet worden. Wer Interesse an einer Mitarbeit hat (ob aktiv oder passiv), würde ich bitten, sich in dieser Gruppe anzumelden!

I wrote it before in my blog, but once (when i was young and wild) i threw a sales rep out of my office, who tried to get into the account (me) by FUDing their competitor. I was responsible for an project in the range of 10 Million Euro, thus i assume they got to much "Deutsche Mark" signs in their eyes. I had a basic simple rule for conversations with vendors: You have to shine for your self. If you need to point to non-product related weak points, there is a high probability, that the business relation will be an unpleasant ride.

I wouldn´t disect HP or IBM balance sheets at a customer meeting, albeit i would disect peformance claims and hint to quirks.I wont use the unclear future of FuSi directly at a customer. I fight on a technical level ... everything else is for sissys not knowing their shit ...

Using FUD is a good fear detector in my daily business. The amount of spreaded FUD is proportional to the amount of fear. Thus i have to assume that Mr. Zemlin of the Linux Foundation is really afraid of Solaris.

In the article "Linux in 2009: Recession vs. GNU" wants to make a point for Linux again. It´s Zemlin FUD time again. I tend to award him the "Golden FUD catapult 2008" for this stuff and his infamous involvement in this story

At first: When all the commercial products are to expensive for a customer, he could opt in recession times for the free stack at Sun. For example running your systems with Opensolaris, Sun Java Application Server, Mysql or the Sun Web Stack for free without support. And when your company have weathered the recession and your budget isn´t such a sad story, just call the Sales Rep. Or you could use the equivalent to a debian like structure. OpenSolaris 2008.11 with Glassfish V3 with Mysql Community Edition. Recession IT budgets doesn´t equals Linux. But i wrote about that before. I wont write about it again here ...

I was upset by another comment. He states :

Zemlin also sees FOSS as remaining strong on the server, with Linux continuing to be the major player. Most of Linux's growth in 2009, he said, will be "at the expense of Sun Microsystems, which is floundering in its business model right now. People look at Linux, and they say, HP, IBM, Dell, Intel and AMD -- these are collectively not going to go out of business any time soon. Then they look at Sun Microsystems, and they say, 'Whoa! This company has some serious financial difficulties, they have an uncertain future -- that's not a safe bet for me.' Nobody is really growing much, but where there is growth, it's going to in Linux."

I want to dissect this statement.

At first. AMD and Intel are somewhat OS agnostic. They make their money with supporting Windows, not with this niche market called Unix x86 market. Sorry ... the Q1CY2008 x86 server market was 1.9 million systems large. The market for MacOS X based systems was 2.6 systems(perhaps this is the reason, why they got the "65nm chip in a 45nm casing" custom build from Intel for the MacBook Air). Server x86 is a an intersting business for both it it doesn´t pay their bills. Or the other way round: If you don´t like Sun x86, you just can user IBM, HP or Dell as well with Opensolaris x86. And furthermore: IBM, HP or Dell aren´t married to Linux. When the three think that there is money to make with Solaris, they will sell it. And they already sell it. And they will drop Linux faster than you can write "penguin" when they come to the conclusion, that it´s not cost effective to sell and support Linux.

BTW, Mr. Zemlin, Sun has no financial difficulties ... it has a problem with its stock price ... the financial reality and the stock price isn´t really correlated (if it´s correlated at all, i have my doubts about it))

Okay: Opensolaris ... the code is opensourced, the cat is out of the bag. Anybody can build it´s own business on the basics. Just a thought game: Let´s assume Sun would go out of business tomorrow. Without a warning. Jonathan says: "Dear shareholders, here is the money. Let´s call it a day. And thank you for the fish". Now take into consideration, that the installed base of Solaris as large as billions and billions. Business is like political power. It don´t like a vacuum. Allowing everybody to participate in the development is just one side of opening the source. The other side is the fact that you enable everybody to support the code. The whole business model of Red Hat or Suse is largely based on this point. Thus in the theoretical case Sun support for Solaris would disappear, it would take a few days until another company fills this vacuum. There is money to earn. Much money. But this whole discussion is hypothetical: There is no vacuum to fill, as Sun won´t disappear. And Sun wants to earn this money. Anyway: The complete discussion and the idea introduced by Mr. Zemlin is just utter nonsense, classic spreading fear, uncertainty and doubt. FUD 101.

I could make a similar prediction about Linux: In the course of the recession more users of will opt for free distributions like Debian, CentOS and OpenSuSE as their budgets don´t allow them to buy support for every system. The sales of subscriptions will decrease. This will lead to financial pressure to the major vendors of Linux distributions. Because of the single-trick ponyness of both vendors (RedHat and Novell) it put both companies on the verge to oblivion. They will cease to exists of purchased by IBM or HP. Futhermore this will bring problems for the core development of Linux (not the drivers, the real core) as the both companies employ many of the core devlopers.

Doesn´t sound reasonable? Well ... then tell why it´s more rational to assume a company with hundreds of billions installed base, billions of revenue a year, positive cash flow, 2 billions at the bank, almost no debts should go out of business or purchased. The last thing is extra ridiclious. It´s credit crunch time at the moment. The hedge fund locusts don´t get money to do they business at all even for smaller deals, everybody keeps it´s money to have liquid money in case of a longer credit crunch phase and we didn´t talked about the anti-trust regulations at all.

At the end Mr. Zemlin just spreads FUD. Large amounts of it. Dear Mr. Zemlin, we can talk about the importance of features like in your last FUD marketing attempt, we could talk about technical advantages of Linux and Solaris. But dear Mr. Zemlin, stop to discredit yourself as a FUD thrower ... otherwise you are burden for the community you represent ... again ...

With gcore you can create core dumps of processes without the need to stop them. You can do the same with the crash dumps as well. You can trigger a crash dump without rebooting the system. The command for this task is the savecore command. The normal task of this program is to take the content of the crash dump device and to create files in the regular filesystem to make them persistent for further studies after the reboot of a system.

After some comments from readers Marina and I updated the article about pfexec at the Sun Developer Network. We integrated some clarifications and changed the wording in some parts.

It`s difficult to write a text, when the jargon is a little bit ambigous. After a writing "root capabilities" several times, you want to use another word ... as i think of having root rights as a privilege (you are a trusted and knowledgeable user), i like the phrase "root privileges". But as privileges are already a common phrase in Solaris 10, it´s a little bit problematic as the capabilities of root are the result of a set of privileges, not a single privilege. So there is no "root privilege" (set this privilege and you are root) in the sense of the least privileges concept. At the end we opted for leaving the phrase "root privilege" in the text at some parts.

Glenn Brunette and Jeff Victor published an interesting document about Solaris Zones with the Blueprint "Understanding the Security Capabilities of Solaris Zones Software":

So why does the world need yet another article about Solaris Zones? Simple. Most publications and sites focus on the consolidation benefits of Solaris Zones. While server and service consolidation is a key use case for Solaris Zones, there is so much more to the technology. Other materials focus on system administration practices related to configuration, installation, management, and troubleshooting. This is incredibly useful information, but there is still an important gap. Namely, many people do not have a full appreciation of the security benefits enabled by Solaris Zones, and sparse root zone configurations more specifically.

Definitely a must-read document!

A really positive article about Opensolaris 2008.11. Steven Lawson writes in OpenSolaris 2008.11: Its Time Is Coming:

I was attracted to OpenSolaris 2008.11 in the first place by a couple of other internet articles, one of which posed a question, the other a bold statement.
The question was: "Will OpenSolaris 2008.11 attract Linux users?"
And the statement was: "OpenSolaris tackles Ubuntu dominance".
Let me deal with the question first by answering simply, 'Yes, if they have any sense'. And as for the statement, my reply is: 'No, it doesn't, not yet - but one day soon it should'.

and

OpenSolaris 2008.11 is well worth investing some time and disk space on. I think we will be hearing a lot more about this operating system in the future.

Of course he sees room for improvement, but at the end the improvments are just "We work on that" issues: More packages (in this article multimedia applications) and more wireless drivers.

I found two really interesting articles about the networking in Solaris: The first one - Examining Large Segment Offload (LSO) in the Solaris Networking Stack - examines the effect of large segment offloading to the cpu utilisation. The second one - CPU Utilization in networking - the TCP transmit path discusses how the cpu utilisation changes with different blocksizes and number of connections.

Jim Grisanzio wrote an excellent piece about the alleged areas of conflict between Linux and Solaris in "It´s still too late":

So, is it too late to catch Linux? I'm not sure it matters much in that context. It's a big world, and there is room for all of us to fit. We simply have too much work to do learning from Linux in some areas where they are strong, focusing on some of our clear advantages in other areas, transforming the existing Solaris base into an open community, and reaching out to new users and developers who have never even heard of us. It's not too late. Not by a long shot. I just don't view projects from that perspective.

I share this opinion. This is the very reason why i don´t like the question "Sun is the bad guy because they have their own Unix and don´t work at the Linux kernel". I´m a Solaris supporter ... but there is room for both operating environments.

Brendan Gregg published an follow-on article to his 250.000 IOPS article. In "Up to 2 Gbytes/sec NFS" he shows how far he can drive the Sun Storage 7410 in a single-node configuration. I like his approach to talk about an benchmark-special result at first, to show it´s pitfalls afterwards just to present a more realistic number. As in the last test, he factored out the harddisks by using a workingset within the size of the main memory (100 GB on a 128GB system)

He was able to get up to 2 GByte/s from a 7410 with two 10 GBit/s interfaces and 20 clients. (with some upside potential, as the CPUs weren´t loaded). In a more realistic test he was able to yield a little bit more than 1 GByte/s over a single 10 GBit/s interface.

That's 1.07 Gbytes/sec outbound. This includes the network headers, so the NFS payload throughput will be a little less. As a sanity check, we can see from the first screenshot x-axis that the test ran from 03:47:40 to about 03:48:30. We know that 50 Gbytes of total payload was moved over NFS (the shares were mounted before the run, so no client caching), so if this took 50 seconds - our average payload throughput would be about 1 Gbyte/sec. This fits.

10 GbE should peak at about 1.164 Gbyte/sec (converting gigabits to gibibytes) per direction, so this test reaching 1.07 Gbytes/sec outbound is a 92% utilization for the 7410's 10 GbE interface. Each of the 10 client's 1 GbE interface would be equally busy. This is a great result for such a simple test - everything is doing what it is supposed to.

An excellent result!

Interesting conclusion of Charlie Schluting in his article "Will OpenSolaris 2008.11 Attract Linux Users?":

If OpenSolaris spent the next 1-2 years focusing primarily on adopting standardized open source software rather than implementing shiny new features, the project would stand a much better chance of attracting long-time Linux users.

Don´t think so. The new features are the stuff that attacts Linux users ... you have to do both. Making it easy for Linux users to choose their favourite OS on technical merits than on usage habits. But there has to be a technological lead. Otherwise there wouldn´t be any reason to think about changing. Often good-enough is good-enough. You need such features like dtrace or zfs or crossbow to create a demand.

OpenSolaris doesn’t need the exclusive community that Solaris enjoyed all these years, it needs the huge Linux community. I think it will happen; the question is, “how quickly?”

ACK. I think Solaris will get more influence from the Linux side. We need the bigger linux community. The problem is: You can´t get a linux-like standard on one side without violating some standards and alienating Solaris users. Solaris don´t need the community at the price of its soul. We have to balance this.

PS: Anonymous is correct ... i´ve read this article in an incorrect way as i was distracted, so it led me to a wrong comment. I´ve reread the article again and edited the article.

After the integration of the first bits of Crossbow, Sunay found some time write about some of the advantages of this network stack: Crossbow - Network Virtualization Architecture Comes to Life.

At 9am PST on opensolaris.com.

The .iso´s for OpenSolaris 2008.11 are available for download now for a few days, but today we had the official announcement: Sun Microsystems Launches Latest Version of OpenSolaris; Unveils Time Slider Visualization Tool and New Partnership with Toshiba to Build Line of OpenSolaris Laptops.