I’m still alive

… but not OCP. Yes in the last times i’ve been very busy and very lazy. So i’ve not written nothing here, i’m still studying (?) to get OCP but work does not leave me so much time to study and to write here something useful or interesting. Now i write this useless post only to say that i’m here and i’m still working with Oracle. by the way, one custemer of my company as recently upgraded from 9iR2 to 11g. That’s incredible but it seems that 11g works fine.


English Dictionary

I’ve been busy for a while, so i’ve haven’t written new posts here. But two days ago i’ve bought a new fantastic English Dictionary: my battle with English language is very hard but i’ll never give up, please forgive me for this. In the meanwhile i still write on my original blog.

About NAS and SAN

I would thank Kevin Closson for the insights the he gives on his blog. Recently i partecipated on a thread on comp.databases.oracle.server on a debate on Oracle ASM over NFS. After reading some “old” posts by Kevin Closson, starting from this and following all the links , it’s more clear to me what does it mean Oracle over NFS, that is Oracle with NAS and what are the advantages of using NAS over SAN. I’ve to say that Kevin refers to systems more complex than those with wich i normally work. He spokes about RAC systems with 10, 25, 50 100 nodes. I’ve never heard here in Italy about such systems.

There is no doubt that with such numbers manageability of NAS is superior of that of SAN, you save all FC cabling  and related costs. On the other hand i’ve found difficult to think about a datacenter with such numbers (clusters with tens of nodes) and only one type of storage type for all (NAS). They (the  datacenters) are frequently more like a mess of old and new technologies so i think there is no hope to reach such unified architecture.

I found interesting a post where Kevin Closson refers to a document of EMC where is a comparation of performance between Oracle with ASM on RAW devices, Oracle on NFS , Oracle with CFS and others (what about iSCSI?). I found interesting the observation that performance with ASM and direct I/O CFS are more or less the same.

RAC Virtual IP: Correction

I’ve studied. Yes i was not completely correct in my previous post on RAC Virtual IP because i’m guilty of not RTFM. It appens, there are thousand of manual pages. So, in Oracle® Real Application Clusters Administrator’s Guide release 10.1 in appendix B i’ve checked SRVCTL command sintax and i’ve found that with “srvctl modify nodeapps” command you can do it. With option “-A <new_vip_address>”  where    The node level VIP address (<name|ip>/netmask[/if1[|if2|…]]). last, optional is the specification of operating system network interface. So maybe that with graphical interface of VIPCA that is automatically started during initial setup of RAC you cannot specify the interface, but you can change it after. So i think that is why the DBA is yet required.

I’ve not tested such modification because i’ve not a RAC test environment. Some time ago i’ve made some tests with vmware virtual machines but those machines are gone away so i’ve to found time to redo that work and made all the tests that i’ve to do

Execution Plans differents with different Users (?)

On google group on Oracle server (http://groups.google.com/group/comp.databases.oracle.server) i’ve seen this:

By accident, I found the solution for this problem on MetaLink :
Bug 4652274 – Explain Plan Differs With Different Users
It has to do with the init parameter secure_view_merging, which is new since
10gR2. Setting it to FALSE in the spfile and boucing the instance resolved
all my problems … Now queries on view from another schema have the same
exection plans when executed by the view owner compared to another user.

I’ve noticed this question because last person to write was Jonathan Lewis. The problem arised seems bad, Oracle in his note says that’s not a bug but an expected behaviour. Since my memory is not so good i’ve decided immediatly to write this here.

P.S. (20/08/2007)

See comments for correct parameter name, thanks to David Jeffrey

Oracle DBA

From oracle-l mailing list i’ve extracted this post:

Not a guru, but I think one of the most important things for a DBA to
know is how to learn – quickly, and just as important for long term
success is to have a natural desire or drive to learn. Databases and
all the interrelated technologies change so fast that you have to be
willing and able to constantly learn, and to be happy with life as a
DBA, you have to enjoy that challenge. By knowing how to learn, I mean
being able to quickly identify what it is you need to know, where to go
to get the necessary knowledge, how to quickly sort through all the BS
to zero in on the key concepts that you need, and then how to apply the
knowledge to your specific situation. There certainly isn’t any fixed
set of topics that a DBA needs to know. The label “DBA” describes many
different roles in real life and in my particular position, knowing C
doesn’t really do me any good because I never look at C code. Rather
than knowing any specific language, it is more important to understand
the concepts of coding and then you can take that skill and pick up the
specifics of any language as needed. Obviously you need to be
comfortable with the SQL syntax and at least familiar with the
procedural code for the RDBMS you are working on. You just have to be
as intimate as you can with all the pieces of whatever environment you
find yourself responsible for. It is a huge plus to have as much
understanding as you can of the operating system and all applications
running against your databases, as well as anything else running on the
same server. Even an understanding of the hardware and network you are
running on can be very helpful. A good understanding of the business
you are supporting is always useful too. I think you really have to be
a jack of all trades and master of at least one in order to be a really
good DBA.

Regarding where to start – there are many paths to enlightenment 🙂 so
just take your pick. Some start as application admins (like me), some
as sys (OS) admins, some as developers – all tend to end up with
different strengths and weaknesses but I don’t think one is inherently
better than the others, just different. Whichever path you pick, just
try to pick up as much of the others as you can along the way

In response to question: “What about non Oracle things we need to know to be a DBA? ”

I like very much the answer, it fits to me and i agree all what the poster has written. I came from a course of studies in information technology that has teached how to do some basic things, then i’ve learned that is most important to now where to start,  we can’t know everything so organization is the basis.

Oracle Database Standard Edition

I will start  saying that my experience on Oracle is essentially with Standard Edition and with features included with that economic version. Partioned Tables, Streams, Advanced Replication, Data Guard are all interesting features of Oracle, but they are expensive related with target business of my company, so i’ve not worked with such features yet and i’ve nothing interesting to say about.

Hovewer administering and developing on an Oracle Database Standard Edition is not so easy. Also a backup for a person who has never seen one in Oracle is not  an easy task (especially if you need such backup to restore a lost database).

My “Oracle” (with the meaning of source of wisdom”) is Thomas Kyte and his last book “Expert Oracle Architecture”