Sun/Oracle

You may think that this already there, but so far you was only able to set a single dataset (ZFS filesystem or ZFS emulated volume) into read-only. This PSARC case proposed such an mode-of-operation for the complete zpool. According to the documentation of this case, the following steps are done to reach the read-only-ness.

1). pool is loaded but transaction processing is disabled
2). all filesystems and zvols are mounted in read-only mode
3). any intent-log replays are deferred (any pending synchronous writes will be replayed once the pool is imported read-write)

Something is going on. The following announcement of an announcement appeared at osol-discuss:

A number of the community leaders from the OpenSolaris community have
been working quietly together on a new effort called Illumos, and we're
just about ready to fully disclose our work to, and invite the general
participation of, the general public.

We believe that everyone who is interested in OpenSolaris should be
interested in what we have to say, and so we invite the entire
OpenSolaris community to join us for a presentation on at 1PM EDT on
August 3, 2010.

You can find out the full details of how to listen in to our conference,
or attend in person (we will be announcing from New York City) by
visiting http://www.illumos.org/announce (The final details shall be
posted there not later than 1PM EDT Aug 1, 2010.)

We look forward to seeing you there!

- Garrett D'Amore & the rest of the Illumos Cast

Really looking forward to this announcement

Bryan wrote an interesting article about his move to Joyent:

Add it all up — the history in the cloud space, the disposition to solving tough cloud problems that I want to solve like instrumentation and observability, and the exciting development of node.js — and you have a company in Joyent that I believe could be the next great systems company and I’m deeply honored (and incredibly excited) to be a part of it!

Last night a really large putback found it's way into the codebase: It centers around the Fault Management Architecture of Solaris. A lot of new features found their way in Opensolaris. For example smtp-notify:

smtp-notify is a small, lightweight daemon that is fully managed by the Service Management Facility (SMF). It uses the interfaces delivered in libfmevent to subscribe to both software and FMA problem lifecycle events. Upon receipt of an event, it produces an email notification based on a set of notification preferences which are stored in the SMF service configuration repository.

or snmp-notify

snmp-notify will generate one of two types of SNMP traps, based on the event class. For FMA events (list.* events), snmp-notify will generate the existing sunFmProblemTrap trap as defined in /etc/net-snmp/snmp/mibs/SUN-FM-MIB.mib. For software events (swevent.*), snmp-notify will generate a sunSweventTrap trap as defined in /etc/net-snmp/snmp/mibs/SUN-SWEVENT-MIB.mib.

Furthermore there are lot of other changes, for example a way to configure the notifications in the SMF repository or the definition of FMA events for instance state transitions (eg. a service going online or offline or in maintainance state).

As SMF is running in each zone on it's own, the fmd (the fault management daemon) had to be enabled to run in non-global zones as well. This was introduced by PSARC 2010/225:

This RFE will enable fmd and associated SMF services in a non-global Solaris zone. The set of fmd plugins delivered into a zone will be restricted to those that are relevant - the hardware-oriented modules, particularly, will not be delivered in a non-global zone.

The first consumer of the fmd service in a non-global Solaris zone will be SMF; this will work with fmd to achieve snmp and email notifications of instance state transitions.

Yesterdays announcement regarding support subscriptions on HP and Dell x86 system has a much broader background. In this article i want to share this information with you:

• Albeit the recent announcement just talked about HP and Dell, this offer is valid for all certified systems on the HCL. There are 508 server systems right now on the HCL. You will find the HCL here. If the systems isn't on the HCL, ask your vendor to certify it. So you are able to buy support and for Fujitsu Systems as well as for IBM systems for example.
• Pricing is socket-based. You pay $1000 for Oracle Premier Support for Operating Systems per year per socket for systems with one to four sockets and $2000 per socket per year for systems with more than five sockets.
• As most of my readers are interested in Solaris, i think i know the next question: As you may know, the Software License Agreement for Solaris for example just allows you to use Solaris 10 up to 90 days in production use without entitlement. This support offering provides you a non-perpetual entitlement to run Solaris 10 on a non-Oracle HW. To say it simple: It's a subscription. This entitlement is valid for the same period as your support. When you don't renew the support, you don't have an entitlement to use Solaris in production after your support expires.
• The support is provided directly from Oracle.
• The rules with Sun servers are different: You get a bundled perpetual license with your server and pay only 8% (SW) or 12% (SW+HW) of the net systems price for support.
• Albeit you have now the choice to run Solaris on non-Oracle x86 servers, i see several advantages to use Solaris on Sun equipment. Those advantages range from "Same vendor of OS and HW" to "Better integration of the system into the Fault Management Architecture of Solaris"

(Safe Harbour: Keep in mind, that this blog is a private one. I don't write in my function as a Oracle employee. The interpretation of the rules expressed by an Oracle representant is the authoritative one)

Just choked on my tea and spew it halfway to my display while reading this:



Okay, i won't call TPM by his other name, as he didn't called me SunHead or so ...

Oracle just announced:

• Oracle today announced Dell and HP will certify and resell Oracle Solaris, Oracle Enterprise Linux and Oracle VM on their respective x86 platforms.
• Customers will have full access to Oracle’s Premier Support for Oracle Solaris, Oracle Enterprise Linux and Oracle VM running on Dell and HP servers. This will enable fast and accurate issue resolution and reduced risk in a company’s operating environment

Richard Fichera of Forrester writes in "Does SPARC Have a Future?":

Our net take is that while it may still be hard to convince a new customer to jump on board the SPARC bandwagon, existing customers have no reason to make an abrupt migration, and can wait and see if Sun delivers on the milestones that they will reveal over the remainder of 2010 and into 2011. As these milestones are disclosed we will have further analysis.

Interestingly there is an comment about Solaris on x86 at the end of the article:

On a related topic which we will address in more depth in future posts, Solaris on x86 will remain a solid enterprise choice for both current and new users. Customer issues with Sun were centered around SPARC and the strategic future of the company as opposed to technical issues with Solaris functionality, performance or stability. The continued availability of this proven operating environment, with an anticipation of continued investment and improvement by Oracle, is a strong positive for any customer looking for an enterprise-class Unix environment on an x86 platform.

The Opensolaris SATA subsystem got TRIM support with this putback to the codebase. The TRIM command is really interesting for SSDs, as the operating system can tell the SSD which blocks aren't used by the filesystem any longer with this comment. This helps the internal garbage collection of the drives to work more efficiently at erasing blocks ahead of their usage. As erasing is really an costly operating with Flash-SSD, erasing the blocks before you need them speed up things a lot.

Habe heute vormittag die Mitteilung bekommen, das mein Vortrag auf der DOAG 2010 angenommen worden ist. Damit werde ich dort einen Vortrag zum Thema "Performance Analyse - oder: Was macht eigentlich mein Solaris?" halten. Die DOAG Konferenz 2010 findet von 16. bis 18. November 2010 in Nürnberg statt.

At docs.sun.com you will now find the docs for the X4800, the X4470, the X4270M2 and the X4170M2. I prefer the the manuals over the slide decks

In a discussion at eWeek a commenter wrote, that the world of IT news doesn't just consists out of datasheet. Yes he is right, it contains reading manuals and to take press releases with a large grain of salt.

Oracle upgraded the Oracle SL8500 library to house 100.000 tapes. This is done by linking 10 libraries together with 10.000 tapes each. This is the largest library you can purchase at the moment. More information is available in the data sheet.

Of course the Chris Mellor of TheRegister can't let Oracle get away with this and points in "Oracle unleashes robo-tapeswapping monster" to the T-Finity library of Spectralogic, and indeed the press release of Spectralogic talks about 120.000 tapes in 4 linked tape library with something called "Skyway".

But there is just one kink with that: The datasheet says "future", the promotional video says "future" and there is no manual (1st rule of IT: When you really want to know something about a device before buying, read the manual, not the datasheet) available at the Spectralogic website. Of course you can't find the word "future" in the press release.

BTW: The Skyway looks like a concept i want to see working for a while before believing that it will work reasonably problem-free. The T-finity complex housing that large number of the tapes consists out of 4 libraries that are linked at both ends by something called skyways. So a tape going from one library to another has to be transported to the ends of the library (sounds unproblematic), has to be put into a device at the top of the connecting racks that builds the T-finity (sounds relatively unproblematic), then it has to be put into the gadget called "skyway", which connects the rows of racks by spanning over the aisle between the tape racks. And that sounds not unproblematic at all to me. In theory this sounds like a nice idea, but i would like to read the manual to see, how they circumvented all the problems i can think of. However .... there is no manual available.

But when we talk about future numbers, i want to cite from the users manual of the SL8500 on page 19:

31 libraries for Automated Cartridge System Library Software (ACSLS) connection, connected through PTPs, containing over 320,000 cartridge slots and 1,984 tape drives.

But that's a "Mine is longer/mine are bigger" contest on a scale totally irrelevant for us mere mortals ...

Bo Steven Zhou proposed an interesting feature for Solaris in the "PSARC 2010/296: Add tunable to control RMW for Flash Devices".

The register writes "IBM splits with Solaris 10 on x64" and expresses, that this is the second vendor after HP. Albeit it was published after the announcement of HP to resell Oracle Solaris licenses, it seems, that TPM wasn't aware of the newest developments . Bets anyone?

An interesting ARC case found it's way into the case log: Stephen Lawrence proposed the zonestat to the architecture review comitee.

This feature consists out of two components. The zonestatd and the zonestat command. Stephen writes about the zonestatd:

A new smf service svc:/system/zones-monitoring:default will execute a zones monitoring daemon. This daemon will implement monitoring the system configuration, managing extended accounting, and computing memory utilization on behalf of clients.

This zonestatd is queried by the zonestat command to deliver data about the resource utilization of zones.

# zonestat 5 1
    SUMMARY
                        -----CPU------------- ----PHYSICAL--- ----VIRTUAL----
                   ZONE USED %PART %CAP %SHRU USED  PCT  %CAP  USED  PCT %CAP
                [total] 9.74   30%    -     - 7576M  23%    - 11.6G  24%    -
               [system] 0.28  0.8%    -     - 6535M  19%    - 10.4G  21%    -
                 global 9.10   28%    -     -  272M 0.8%    -  366M 0.7%    -
              kodiak-ab 0.32  1.0%    -     -  256M 0.7%    -  265M 0.5%    -
             kodiak-rie 0.00  0.0%    -     - 22.7M 0.0%    - 49.4M 0.1%    -

Due to the implementation with a daemonized part, you can even tell the daemon to compute a report about the utilization in certain intervals:

# zonestat -p -r memory -q -R average,high 30s 24h 1h

Please read Stephens case document for for more information.