No Huddle Offense

"Individual commitment to a group effort-that is what makes a team work, a company work, a society work, a civilization work."

Keep your OpenSolaris installation slim/clean

September 18th, 2009 • 3 Comments

OpenSolarisLogo2I really like OpenSolaris. If you haven’t tried it yet you should give it a try. Still there are some things where I choose other operating system over OpenSolaris. Missing ports for some of the application I use is one point…

But never the less OpenSolaris has some killer features which makes is worth a try. For example ZFS, Zones or DTrace.

One tool I absolutely like when using Linux is the packaging system and the ability to clean up orphaned (e.g. with help of deborphan) packages. After some googleing it seemed that there is nothing in OpenSolaris yet to do so. But here are some tips and tricks to keep your installation clean as long as the feature is not available…

  1. On a laptop use ZFS compression:
    zfs set compression=on rpool
  2. Turn of the caching in pkg. (This will save you a lot disc space!)
    pkg set-property flush-content-cache-on-success True
    Downloaded packages are then no longer kept on your harddrive in /var/pkg/download
  3. deactivate unneeded services / uninstall software. This is a personal one – Decide what tools you wanna have and which ones not (Presumably you can uninstall a lot of the languages packs). Some service which you could deactivate are:
    svc:/application/desktop-cache/input-method-cache:default
    svc:/application/desktop-cache/mime-types-cache:default
    svc:/application/desktop-cache/icon-cache:default
    svc:/application/desktop-cache/desktop-mime-cache:default
    svc:/application/print/ppd-cache-update:default

    Use the following command to do so
    svcadm disable ${svc}
    svccfg delete ${svc}
  4. Find leaf/orphaned packages – This is more complicated but also a very important part. Especially some libs which you do not need any longer tend to stay. What you can do is the following – Find all the dependencies using
    /bin/pkg search -l \'depend::\'
    And then find all installed packages:
    pkg list
    All items which are installed but nobody depends on – are therefore leafs/orphaned. But be carefully – this will also list eclipse etc. But this is a way to find the orphaned libs. Personally I did some parsing with python to parse the two list of dependencies and installed packages to find the unique ones.

Overall these actions saved me 4Gb of space. Next thing to play with is tuning ZFS 🙂 Installing it on hybrid system. Have a look here: http://blogs.sun.com/brendan/entry/test

Haiku OS Alpha released

September 14th, 2009 • Comments Off on Haiku OS Alpha released

Haiku-logoFor those who still know BeOS – Haiku OS released a Alpha version! http://www.haiku-os.org

Sun HPC Workshop

September 14th, 2009 • Comments Off on Sun HPC Workshop

Last weeks HPC workshop has been great! Very interesting presentations and a lot of good talks. Nice to meet and great the people behind so many projects inside and outside of Sun working on Cloud, HPC, Sun Grid Engine, Service Domain Manager and similar! Slides can be found here: http://wikis.sun.com/display/SunHPC09. This includes my slides: http://wikis.sun.com/download/attachments/170755116/20090825_open_cloud_frameworks_hpc_tmetsch-v0.2.pdf

For Your Cloud Poster

September 7th, 2009 • Comments Off on For Your Cloud Poster

So happy to see the poster I made at the Sun HPC workshop:
28285391

Open Cloud Framework – Open Standards for the Cloud Community

September 2nd, 2009 • Comments Off on Open Cloud Framework – Open Standards for the Cloud Community

Here is the talk I gave at GridKa School 2009 in Karlsruhe. Recordings will follow later – so stay tuned. The slides itself might not be to easy to read because they are designed for presentation not for Offline usage. But you’ll find a short transcript below…

  1. Introduction slide
  2. A very simple quote showing that not many people know what Cloud Computing is. Simon Wardley – found 68 definitions at his OSCON presentation.
  3. So what is cloud computing – cartoon shows that it is unclear and that Cloud Computing is not when you take an existing product and add a ‘Cloud Computing’ stamp on it. But these slides will define Cloud Computing using the RESERVOIR project acronym. The cartoon although is funny – the first line shows that adding public resources to your Private Cloud will create a big cloud. The second line shows that if you deinstall all software you will get Sun hardware 🙂 The third shows that if you have problem with your Cloud Sun will fix it.
  4. Intro to the FP7 EU Project RESERVOIR
  5. RESERVOIR stands for Resources and Services Virtualization without Barriers. So the 3 components are Resources, Services and Virtualization. If you combine them all without barriers you get Cloud Computing.
  6. Resources can be one of these nice Sun Blade Centers
  7. Services are any kind of Software with a demand for Services. The last think is important. Without the demand no Cloud. Next to that a Service has an Interface towards the End-user a Description consisting of Meta-information for the Semantic Cloud as well as requirements like CPU speed and architecture. Also a Service needs to have a SLA bound to it.
  8. Virtualization is basically an abstraction. And possible for Hardware (VirtualBox, VmWare,…), Software (Java), Storage (OGSA-DAI) or Network (Sun Crossbow).
  9. A lot of barriers need to dealt with.
  10. Sun doesn’t do this alone but with these RESERVOIR Project partners.
  11. Still what is Cloud Computing? This picture shows that it is everything and a kitchen sink. It is a lot and therefore we need more clarification based on Classification.
  12. The 3 major things which can be provided as a Service towards the customer are Software, Infrastructure and Platform as a Service. Important is that this is not a layered approach because you could offer SaaS on top of IaaS without PaaS…
  13. Several patterns of how users use the Cloud also show that Cloud Computing isn’t easy to describe
  14. The diversity also adds more confusion
  15. But one major thing of the Cloud is that you can do migration, consolidation and Hybrid-Clouds – so to say beeing elastic
  16. The RESERVOIR layered architecture
  17. Demoing the features fo such an environment – things Sun has demonstrated
  18. The Virtual Java Service Container is a unique entry point for all kind of Java Services. Those can be deployed and then managed/scaled in the environment
  19. The Exeds of a SGE cluster can be started and shutdown on demand.
  20. But the ‘blue arrows on slide #16 show that all this (elasticity, management etc.) is not possible without Standards. One is OGFs Open Cloud Computing Interface.
  21. A nice quote about standardization – design it to be extensible and think about the future
  22. It has 3 main drivers: Interoperability
  23. Portability
  24. Integration
  25. Focus on IaaS/virtual workloads
  26. If you have a lot of people you will have also some great minds onboard who can help
  27. 190 Members and a lot of active members/implementors and chairs are helping out
  28. The timeline to create a slim, extensbile API for the Cloud (And one of the first!) by OGF27
  29. The two devliverables of the OCCI working group
  30. The CRUD operations and mapping towards HTTP operations
  31. A diagram of OCCI. Maybe to complex to explain in text. Join the group or listen to one of the future talks about OCCI 🙂
  32. The most important slide. Having only one standard doesn’t help. It helps having one interface to ‘combine’ Clouds but what happens with e.g. secure data (top left)? So there is a demand for other standards. Those are shown in the big Cloud. More are also there. So please: join and help those groups! Without these the Cloud will not be possible (think of the barriers and elasticity)
  33. If it doesn’t happen Mankind will also die off…
  34. References and questions