2007/08/29

back to the SAN

yeah, I should have done this a long time ago.


Unfortunately, been busy as heck with a Peoplesoft upgrade that somehow degenerated into installing and setting up two additional servers and a load balancer with the usual plethora of problems, compatibility glitches, version clashes, appserver performance issues, etcetc...


Had to put my foot down and explain very clearly that my field is databases and data management and none of the above has anything to do with what I'm trained to do! Some people have difficulty in understanding this.


I also have a problem with project "plans" that get ammended every day. Frankly, that's not a plan: that's a diary!...



But, I digress...



Kevin blogged recently for a while on SANs and NAS/NFS and their many relationships and uses with database servers.


Apparently, it all started with a comment I made here a while ago about how easy it is to reconfigure a storage facility with a SAN.


Some folks assumed I was in some form criticizing Kevin's many interventions on NAS/NFS and its advantages compared to SANs in the database server universe.


Now, let me be very clear about this: in the last few years I've been discussing storage technologies and exchanging opinions with Kevin through the oracle-l list first and later through the blogsphere - and many times through email which he's always been gracious to reply to - and I have yet to see one instance where we disagree in a major way.


Sure: our experiences in the storage area may be slightly different in many cases and that certainly explains some of our small differences of opinion. But by and large, I can state that if I ever get stuck with an I/O capacity problem, I need only one person helping me and that is Kevin.


He knows his I/O in detail and in depth and I agree with almost every single one of his well researched and accurate statements and opinions. As anyone following some of my replies in his blog would probably have found out by now.


As such, it was with surprise I sensed folks thought my comment a while ago was in some form derogatory or disparaging of anything he had said.


Let me be very clear: that was most certainly not my intention!


What I may have failed to explain back then is that I tend to prefer SANs for our environment for one major reason: we run a mixed shop with many, many SQL Server and Lotus Notes servers as well as our Oracle dbs. And multiple Wintel file servers and Citrix nodes which all end up needing disk space in some form or other.


In such an environment, it is hard - at this stage of the technology! - to integrate a single NAS service solution.


SANs have been around longer and make these mixed environments easier to configure than for example with server native disks.


They also allow for flexible mixes of the Wintel and Unix universes - in our case AIX and Solaris. "flexible" in that we can easily reconfigure the IO setup between all this cacophony of sometimes conflicting demands for storage.


Moving on:


At the moment we are far from being I/O constrained in the Oracle servers. But if that ever happens, my next port of call will surely be add-on NAS/NFS technology.


And here enters the determining factor in all these musings: "suitability and integration of a technology, within budgetary constraints". Something that is closely related to the often quoted "real life", although I do hesitate in using that old chestnut.


Of course dedicated disks and their respective controllers in a db server, in sufficient quantities, provide the absolute best speed and overall performance. Why else would the TPC folks use them if that was not the case?


The question being: do I have a need for that level of performance within the overall constraints of my budget and environment? Or perhaps more importantly: does my site NEED that level of performance right now?


If so, then there is no question: I have to use that technology. But if not, then other solutions might be more appropriate in their cost effectiveness.


The next one down from native disks in my list of performance preference would definitely be a NAS setup like the ones Kevin talks about. Particularly if I am running a Linux environment. The advantages have already been touched on by him and I won't dwell on that now.


The last solution for me in terms of speed/performance is the SAN. It's also the most flexible when dealing with mixed environments. And in our case at the moment, that is the major factor in a decision about storage technology. So we use a SAN setup. But it is important to understand why.


This is not to say we won't add to it, at some stage, a NAS facility! In fact it is likely we may do so in the next couple of years, as our DW continues to grow and it becomes cheaper to just store the lot than to fight over archival processes.


But that's another battle, for another blog...


Now folks: do yourselves a favour and go read Kevin's blog on configuring lgwr and dbwr in your Oracle databases. Trust me: take a nice long time at it and ask questions on anything you find harder to understand. It's worth it.


Catchyalata, folks!

2007/08/22

10.2.0.3 revisited...

been faffing around with a 10.2.0.3 upgrade in one of our test boxes.


Eventually gave up on it: got to get this box out to the testers and I don't have time to debug a perl script that wasn't finished after 5 hours of solid running!


And then mr oracle complains when I tell them they should get their act together before flights of fancy on new releases!


If nothing else, at the very least the option of installing a clean 10.2.0.3 without this horrible patch script of theirs.


Have you tried to get 10.2.0.3 standard release - no upgrade - for AIX? Or most of the ports?


Thought so.


And this is the latest release on most ports, given that 11g is not out across the board...




Anyways, some fun:



Most folks now at Oracle wouldn't have a clue where this was. Just up the hill from the current Oracle park in SF.


Shortly after that photo was taken in 87, I drove here:





It's called Mt Connor and to my mind it is even more impressive than Uluru - aka Ayers Rock.


A few years later, we also visited this one:



Somewhere in Austria, between Salzburg and Innsbruck.


You probably noticed I've been dusting off the old photo cabinet?




catchyalata, folks!

2007/08/06

just my luck...

when everyone is having fun watching a royal slugfest, while I'm stuck with a Peopletools upgrade for days and days...


Barely catching up with what's been happening in the blog community in the last few weeks. And I still need to go back to Kevin's great series on SANs and NAS/NFS.


Ah well, I'll get there eventually.





It appears Don has once again managed to ruffle Jonathan's feathers. And vicky-the-versa.


No, I won't provide links: sick and tired of Don's old technique of turning any linked page into an ad to his services!
Oldest trick in the book. But I'm older still.


Let's get something very clear here: I haven't been the target of Don's so-called lawyers. Ever. Yet.
But I know for a fact he's tried to find my address details. Not that it tremendously upsets me: they are publicly available. But it requires the application of attention. Something I know Don appears sometimes a bit short of.


I've read and re-read the claims from both sides. Quite frankly: anyone who believes Jonathan is "after" Don needs their head examined.



Why?



Allow me to elaborate with some historical perspective. It goes like this:



- About 15 years ago Oracle noted there was a lack of good technical doco for their software. They then engaged a number of folks to produce lots of books. Some were from the then International User Group, others like Don were just encouraged to publish as much as possible within the guidelines from Oracle: "don't show any problems with the product".


Not necessarily a bad thing:
"some books" is a lot better than "no books".
Ask IBM and their extensive DB2 "library"...


Of course, along with some very good books a lot of dross was produced. Let's ignore that. Most were along the lines of "re-work the manual" into something of lighter digestion. Again: not necessarily a bad thing. Indeed, anyone who has had to reference large tracts of Oracle's doco would welcome a summary of the stuff!


- Then a few years ago, around the 2000 timeframe, a new group of folks started producing very good, nay, excellent technical documentation. The future Oak Table folks mostly but also some from outside that organisation. This stuff was truly good, with a lot of reasoning and investigation put into it.

It considerably raised the bar on technical literature and it replaced the usual "because I say so" authority with much more solid information, the kind that can be verified by anyone. A leap ahead in quality.


- Of course: this was not welcome by the "re-work the manual" and "publish by the kilo" brigades. And the scene was set for a "war".


Why it would have to be so, to this day beats me: IMHO, there is a place for both kinds of books.


Like I said: there is a need for compact doco. Anyone who has had to wade through the 4 recovery manuals plus the Dataguard stuff will welcome Robert Freeman's RMAN book! Same goes for Mike Ault's and Tim Hall's condensed papers on many, many aspects of Oracle administration and system setup.


And of course: there is also a need for the sort of stuff covered in Steve Adam's and Jonathan's books and so many others of the same ilk. Do I need to explain why?



- So, the "book wars" started. Yes, there were some very public bunfights. I followed closely the ones in c.d.o.s., namely with the "reorganize indexes" war.


But it all got too personal and quite frankly, I've got better things to do with my time. At one stage, it even spread to the oracle-l list: I do remember being the target of innuendo there. That was dealt with at the time and I hope it was clear that while I don't feel the slightest inclination to "write books by the kilo", that doesn't mean I'll take nonsense from anyone. Still: not important.


- One thing seems to be a constant of this whole saga: when Don sees anyone critique his statements, he thinks someone is attacking his publishing business and that entitles him to fight back dirty. This involves email threats, innuendo campaigns, "funny" posts in his blog and site, etcetc.


As someone running a publishing business I guess he feels he's entitled to do whatever is necessary to protect his turf.


That is fundamentaly wrong as no one is really attacking the business itself. Just questioning some of his very public statements and claims. Last I checked, Jonathan was not in the slightest interested in running a book publishing empire! But he's a darn good scientist and has the mindset of an engineer when it comes to technical matters, so it's no surprise that there will be clashes there...


And this is where the important bits really come up: I have yet to see Don reply once with complete facts or detailed technical background to anything that Jonathan points out as incorrect. It's really simple: counter with verifiable facts, not "it's like this because I say it".

Nope.

Somehow, the whole thing always degenerates into a site changing exercise, attacks on locked blogs and other such.


I have yet to see Jonathan resort to that sort of tactics. All he does is explain why some of Don's claims are wrong. And of course once Don starts the credibility attacks, it all degenerates into a bunfight. Yet somehow, I still have to see Jonathan publish any direct attacks on Don's publishing company...




Still, the whole enchillada is starting to reach the ears of Oracle's senior marketing. Figure that: it only took them what, 5 years to click into something being wrong? Amazing demonstration of their "pulse" on what's goes on in the Oracle community. Could it be they don't get accurate feedback from their UG folks?
Narh, perish the thought!


Be that as it may, my gut feel is that it will all end up badly. It is obvious that Oracle will feel compelled to do something to stop this state of affairs. But exactly what, will be the interesting thing to watch: after years of tacit approval of Don's business model and the "ACE" program nonsense supporting the new folks, it'll be interesting to see what they come up with. Then again, they might stick their collective heads back in the sand and really hope it'll all go away...



Tickled pink here! :-)



Ah well, I'm glad that for once it's not me on the hot seat with Oracle's "clear-headed" marketing. Oh hang-on: this blog entry will probably endear me once again to them.

Yup.

Let me see: they'll now send another of their boyz to keep me "under control" at the new job?

What else is new? It's only been like that for what, 18 years?
:-)




Anyways, on a lighter note:




It might be difficult to see but check the coffee shop's name on the window! It's a local shop and the rider runs a local school. Every Saturday morning, there he is having his coffee break!



This is from near where I work now:




I like the "disintegrating building" effect!


Catchyalata, folks!