[prev in list] [next in list] [prev in thread] [next in thread] 

List:       opensolaris-sysadmin-discuss
Subject:    [sysadmin-discuss] Update on SysAdmin Community Management
From:       njd () ndietsch ! com (Nathan Dietsch)
Date:       2006-04-15 21:26:29
Message-ID: 4441C76F.8000201 () ndietsch ! com
[Download RAW message or body]

Hello All,

> I'd also like to elect Nathan
>Dietsch to "Availability and Virtualization" topic.
>
Sorry about the delay in responding to this, like most of us, I have 
been very busy. The Easter long weekend has given me a chance to catch 
up on things.

>Here are some action items for the leaders to work on while I'm away on
>vacation:
>
>1. Introduce yourself to the community and the area you are responsible
>for:)
>  
>
My name is Nathan Dietsch and Octave has put me forward for the 
Availability and Virtualisation topic. I am a UNIX Systems Administrator 
and have been working with Solaris since 2.5.1. I am a consultant and am 
currently working with a large ISP in Sydney, Australia. I have a 
Masters degree in Information Technology and my research was focused on 
how Change Management affects the work of System Administrators. I am a 
member of USENIX/SAGE and SAGE-AU.

Back in 2004, Ben Rockwood asked me if I wanted to be part of the pilot 
program for OpenSolaris, I am now glad of having an opportunity to 
actively contribute. Like many people, I have not yet deployed Solaris 
10 into production and am on the steep learning curve.  I see many 
opportunities for some of the technologies introduced in Solaris 10 and 
OpenSolaris to simplify the environment I am currently working in.

The environment I work in is composed of multiple geographically 
separated Veritas clusters running Oracle databases and third-party 
application software for an ISP billing platform. I have also worked in 
Europe for a number of investment banks with similar environments 
(Oracle back-ends and third party application front-ends). I have been 
working with clusters for about 5 years, more VCS than Sun Cluster though.


>2. Start conversations about your topics. 
>  
>
One of the topics that I am personally interested in is how the new 
technologies available in Solaris are able to increase the level of 
abstraction between servers and services. Furthermore, I would like to 
see these technologies used outside of the realms of clusters and being 
used as part of a standard approach to configuring applications on UNIX 
systems ...

"Virtualised by Default"

By this I mean that applications when installed should be independent of 
the servers on which they run. That is you have a virtualised DNS 
service, not a DNS server. That service can run on any of the servers 
deemed fit for the purpose. Virtualised LDAP or mail servers could also 
run on the same group of servers. When applications are virtualised by 
default, the ability to scale them increases. Not only does it help 
scale individual applications, but it takes much of the pain away from 
consolidating or moving them to other servers.

Over the years, clusters have used the service or resource group concept 
to separate the services (application resources) from the servers 
(cluster nodes) that they run on. This gives clusters a level of server 
independence that provides for high availability where applications can 
be migrated between nodes. The same idea has been used in other areas 
such as webfarms, which are virtualised behind load balancers.


Clusters usually provide two main functions;

* Virtualisation
* Availability

Solaris 10 and Open Solaris provide ( or will soon provide ) a number of 
technologies in both of these areas;

Virtualisation

* Zones (obviously)
* Project Crossbow
* Xen

Availability

* SMF (replace service group/resource dependencies)
* SVM and ZFS
* IPMP (high availability for networks) and Project Clearview.

Providing best practice documents for the use of these technologies can 
help in their acceptance in the system administration community at 
large. The Sun blueprints program already provides this, but I believe 
there is room for a series of community-supplied articles on how the 
technologies available in Solaris are being used to solve the problems 
that System Administrators are faced with on a daily basis;

Examples could include, but certainly are not limited to;

* How do I provide high-availability without the cost of implementing a 
cluster?
    - Yes, I know Sun Cluster is a freely available product, but Sun 
Cluster admins are not so free and the added complexity adds a hidden cost.
* Examples of the "Virtualised by default" idea that include the use of 
the technologies listed above. This could include recipes for common 
applications;
    - DNS
    - LDAP
    - Mail
    - Oracle (and other databases)
    - Example third party applications, the concept can easily be 
extended for specific applications
* Integration of the above with Jumpstart to allow for easy deployment.

>3. Brainstorm about what should be added to the SysAdmin community
>page.
>  
>
Another interesting idea, while not specifically limited to the 
Availability and Virtualisation topic, is the Juniper Masters of IT 
podcast series;

http://podtech.net/?cat=34

The idea is that an interviewer speaks to Network Administrators about 
how they have used Various Juniper products in their environment.

I think something like this for how System Administrators are using 
Solaris in their environment would be great. We already have the IO 
Podcast which deals with Podcasting from within Sun for those on the 
outside, in my opinion Podcasting from the outside is just as important 
and is where the rubber hits the road for Solaris (so to speak).

An example of this is Ben Rockwood's podcast page;

http://cuddletech.com/audio/index.shtml

Anyway, that is enough from me.

I hope the OpenSolaris SysAdmin community is a success.

Kind Regards,

Nathan Dietsch

[prev in list] [next in list] [prev in thread] [next in thread]