Yesterday it was announced that Xen support has been putback (“committed” in CVS/SVN speak) to Nevada. The source is available now and the SX:CE release of Build 75 will be coming soon. While we’re talking about Xen, this is being released as “xVM”, for the following reason:

The xVM hypervisor is based on the work of the Xen community. The name
‘Xen’ is a trademark with some restrictions, so we have chosen to use a
separate name to describe Sun’s hypervisor.

This is an exciting announcement and its the culmination of a lot of hard work on the part of Dr. Tim Marsland and the “Matrix Team”. Congrats to everyone involed! This really finally rounds out the Sun virtualization strategy as I see it, with LDOM’s, Containers, and now xVM, as in house innovative efforts, in addition to existing 3rd party support from partners like VMWare.

The Intel Developer Forum is now over, which closes out the years big Bay Area conferences for me. My only regret this year was that there was no OpenSolaris presence at LinuxWorld because I was on vacation during the event.

For those who were planning on seeing me at CEC in Las Vegas, I’m disappointed to say that the organizers rejected my “iSCSI & The Future of Storage” talk and then today canceled the presentation that I was going to co-present on Virtualization. Really, its not I but the show attendees that will miss out on unique real world perspectives… I have plenty to say about CEC but won’t.

And so we now can collectively turn our attention toward the first even OpenSolaris Developers Summit in North America, to happen in mid-October in Santa Cruz, CA. Register today, the event is free and will bring together the biggest names in the world of Solaris, some old and some new, to look toward a bold new future and chart our path to glory. Its going to be a landmark historic event, so do not miss this unique opportunity to be part of it!

There is a interactive Tech Talk with Ian Murdock & Marc Hamilton TONIGHT in Santa Clara (SCA03). Register now and be there! If you see me, stop by and say hello.

Here’s the address info:

When:  Monday, September 17th
Time:  6:00pm

Location:
Sun Microsystems
Bldg: SCA03 - Auditorium
4030 George Sellon Circle
Santa Clara, CA 95054

When we understand to our satisfaction how something works and we can do it quickly we say “Thats easy.” It may or not be easy for anyone else, but in your mind it is. The converse is true, when something just doesn’t make sense or seems to take more work than you think it should we say “Thats hard.” As SysAdmin’s we are eager to say the the former and embarrassed to say the latter. General this embarrassement gives way to frustration and anger, and if we didn’t like whatever it is in the first place we just say that its stupid or poorly designed. Given that, theoretically, a sysadmins job is to use and implement technology, not to re-engineer it, our frustrations usually have more merit than vendors give us credit. The reality is that our jobs are too often about decrypting technology and there are two parties to blame: engineering (for not making it simpler in the first place) and techpubs (for not explaining plainly how to use the damn thing).

Within Solaris this epic struggle between engineering and techpubs is easy to see. My favorite example is network configuration in Solaris… its actually very elegant and well designed, but its counter-intuitive to most people, so bad bad tech pubs. Another example is ndd, its just crap, its difficult to use and easy to make mistakes, so bad bad engineering. Sometimes it comes down in the middle, such as Resource Control… different commands, funky syntax, no way to just dump out “here are my settings!” without digging through piles of man pages and mailing lists.

Given the option of re-engineering it or writing better docs, the latter is much easier to do and less time consuming. And thus, I ask….

What do you feel is “hard” in Solaris? What are those things you think you should understand but never really have?

I’ll admit that I struggled heavily with RBAC, with resource control, with auditing, and honestly I still can’t make any sense out of extended accounting. Even those of us who are called “guru’s” have plenty of weak spots, and that shouldn’t be supprising to anyone, Solaris is a really, really big operating system, as is any UNIX system.

So, have at it… what is too hard? With a good list those of us in the OpenSolaris Docs community can target our efforts to make the world a slightly better place to live and work in.

Solaris storage enthusiasts will be excited to know that the OpenSolaris iSNS Server Project has released the first usable iSNS Server implementation. I personally refer to this release as “v2″ because some of you may recall Sept ’06 iSNS Server Release which was completely unusable, after which the iSNS Server Project went dark for some time. But, all that is behind us now! This release completes the iSCSI trio officially now that we have a usable iSCSI Initiator, iSCSI Target, and iSNS Server all available as CDDL Open Source.

(If your jumping up and down saying “What about COMSTAR!?!”, I’m not commenting on that right now… yes I know about it, yes I’m interested, we’ll talk later.)

The iSNS Server implementation is available now (released 2 days ago, sorry for the lag). You can download drop-in binary tarballs for X86 and SPARC, or to have a look at the code the WebRev is available here. The code is not yet available in the NWS consolidation OpenGrok, I’m hoping it’ll appear soonish.

To get started with the iSNS Server, download the tarball appropriate for your platform and untar from root (/). You’ll get 3 files: an SMF Manifest, the iSNS server binary, and the iSNS admin CLI binary. Import the manifest (“svccfg import …”) and then start the service (“svcadm enable isns_server”) and you’re good to go.

Now that you have an iSNS Server running you can start to play. I’ll blog about iSNS in detail soon now that there is a reason, but here are some quick notes to get you started:

Once the iSNS Server is running configure both your Solaris Targets and your Solaris Initiators at the iSNS Server, here is an example:

First the iSCSI Target system (I had to restart the target service after doing this):

# iscsitadm modify admin --isns-server 10.0.0.17 --isns-access enable
# iscsitadm show admin
iscsitadm:
    Base Directory: Not set
    CHAP Name: Not set
    RADIUS Access: Not set
    RADIUS Server: Not set
    iSNS Access: Enabled
    iSNS Server: 10.0.0.17
    Fast Write ACK: Not set

And now the Initiator:

root@aeon /$ iscsiadm add isns-server 10.0.0.17
root@aeon /$ iscsiadm modify discovery --iSNS enable
root@aeon /$ iscsiadm list discovery
Discovery:
        Static: disabled
        Send Targets: disabled
        iSNS: enabled
root@aeon /$ iscsiadm list isns-server -v
iSNS Server IP Address: 10.0.0.17:3205

Once this is done have a look at your iSNS configuration and you should see your initiator nodes and your targets as part of the Default Discovery Domain:

 isnsadm list-node -v
iSCSI Name: iqn.1998-07.com.cuddletech:aeon
        Alias: aeon
        Type: Initiator
        Network Entity: iqn.1998-07.com.cuddletech:aeon
        Portal: 10.0.0.2:61191
                Portal Group: 1
        DD Name: Default
iSCSI Name: iqn.1986-03.com.sun:02:97e538ed-f9c0-6c11-9326-eae82f4f1174
        Alias: pool/local/default/aeon
        Type: Target
        Network Entity: flux
        Portal: 8.12.35.49:3260
                Portal Group: 1
        DD Name: Default

That should start you on your way. More on iSNS to come, I want to discuss it in much greater detail.

To the iSNS Server Project developers, namely Hyon Kim and Victor Li. Great work guys!

I’m was greatly saddened to read this: NetApp Sues Sun for ZFS Patent Infringement. I’ve been a big consumer of NetApp and Sun equipment for years. Readers of this blog have read numberous articles discussing the use of iSCSI and ZFS on NetApp in a Solaris environment.

On a business level, I get the reality of this. Sun’s lawers foolishly walked into a bear cave and the bear woke up. -1 to Sun’s lawers. They have no real recourse except to protect themselves and return the favor. If Dave’s account is truthful, and I don’t doubt that it is, this is all Sun Legal’s fault. If I were a NetApp investor (I’m not) I would expect the business to respond and if possible turn it in NetApp’s favor.

On a personal level, I’m bummed. NetApp wouldn’t be what it is without NFS. I’ve long said that “NetApp is what it is because it beat Sun at its own game.” For those of us who are passionate in the Sun storage community the shame of watching a competitor consistently kick Sun’s ass with its own creation has been frustrating and embarrassing to some degree. Nevertheless, it can not be ignored that NetApp and the WAFL file system have considerably advanced the craft and set the standard for storage in the industry. I feel the pain of both sides, but protecting patents such as the COW are somewhat absurd. Yes, NetApp has a legal card to play and they are putting it on the table and, to my knowledge, they haven’t tried to play that card so far, but its still lame. Poor form on both sides… but I admit, if I were to choose a side that I think has more merit it wouldn’t be Sun… and I’m really saddened to say that.

Here’s my question…. where does Jonathan fit into the picture and what can he do? Jonathan and Dave are both very reasonable and intelligent men, I wonder what could be accomplished by providing them with a table, 2 pints, and about 2 hours. Its moved beyond that already it would appear, but I hate to see the courts come into this. And, beyond that, Sun’s legal team wouldn’t go after such a large corp as NetApp without the consent of the CEO right? I know that Sun has been careful all along to stear clear of NetApp patents, but regardless, would the money from the patents be worth more than the hit Sun will inflict from its potentially weaker position in the matter?

What will happen, what will happen…. who knows. I hope it goes well. The industry needs head-to-head battle between Sun and NetApp with victory decided in the market, not the courtroom. EMC, Hitatchi, IBM… these guys aren’t bringing innovation to the market like Sun and NetApp have been. And so I’m praying for a Win-Win scenario to play out. I’m not sure what that is, they are competitors, but I don’t see good coming out of the legal struggle for either party.

Dave and Jonathan, Good luck to you both.