Our DBAs said support of 11g is at least a year off. The general attitude is they aren’t interested in the new features and do not want to learn them. They seem unwilling to support new features in the database and often have to have new features explained to them (in 10g, which has been out awhile) for the first time. They are still using PFILEs in 10g, which is not best practices 10g.
Brian Fedorko |
It always seems that the innovations that truly matter are the ones that impact the bottom line.
Transparent compression will save tangible amounts of money in disks, tapes, comm – The savings increasingly scale with database size.
‘Usable’ standby databases take the sting out of hemmoraging maintenence costs ‘just in case’ and provide for much more ROI.
Better security features, coupled with encryption will save capital on litigation defense.
Honestly, these points don’t seem to be pushed very hard, as the benefits are quite obvious. Features such as Database Replay are receiving a lot of fanfare as they seem much more likely to collect dust. It is a handy sounding idea, but when would you use it?
The only times companies have fresh new databases for testing purposes are generally after making a hardware purchase (New business areas or upgrades). Hardware Vendors generally don’t let you try before you buy – You buy the best you can afford with the budget you’re alloted. What will you gain from testing with Database Replay that you wouldn’t get from making the system available?
I suppose it would be useful if you were downgrading equipment, but who does that?
In summary – Features that make the database smaller, cheaper, and faster will ALWAYS be popular. Features that may seem handy in certain, specific situations? Not so much.
Charles Schultz |
When Oracle came out with 10gR2, we quickly upgraded our 9i ERP thinking that surely, 10gR2 would be that much more stable than 10gR1. Unfortunately, we spent a lot of time on the phone with Oracle Support talking to analysts who were still figuring things out for themselves. During the process, we learned to turn off several new “automatic” features such as Automatic Memory Management and Statistic collections. We did end up taking advantage of other new toys like the ADDM and AWR, but it took a while to understand what they were doing and how best to leverage them. In fact, we are still learning about them. =)
Going from 9i to 10g, in retrospect, was both very painful and very beneficial. Painful because we are caught in an upgrade cycle where we are constantly playing catch-up with new patches and patchsets, not to mention the slew of tauted features that do not fully work. Beneficial because we have learned so much.
With 11g, it seems common wisdom that nobody seriously puts the first release of a major version into Production. Some of the risk-takers and well organized IT shops might (I am sure Dell will), but Oracle has a bad reputation for first releases. We talked to Dell (and others) who beta tested 11g and as Oracle Partners, worked hard to make it solid; we were told that 11g will be one of the most solid first releases in a while. Cynic that I am, I say the proof is in the pudding.
From my perspective, one of the most constant struggles in the decision to upgrade or not is establishing a business reason for it. Face it, most of the new features are not necessarily required by your average shop. They are nice (when they work) and they come with glossy “cool points”, but do they actually fulfill a need or requirement? Like I said, for some shops, possibly yes. I would wager that they are the minority.
I do appreciate that Oracle has actually decided to backport a major feature to 10g. I think that is a good show of confidence. Also, I realize that Oracle’s roadmap for their flagship product does not implement itself overnight, and some features are being matured from version to version. Rolling upgrades and patches are one good example of that.
You asked what the process is with rolling out new features. For 10g, we installed it vanilla and went from there. I have not yet decided how to approach 11g. It makes more sense, in terms of keeping the applications and services stable, do ease into new features. Have we found Oracle helpful in that front? No. At least, not in terms of what we expect of Oracle Support. Over time, I have found that the documentation is absolutely amazing! It takes a long time to not only discover that, but also to dig through the copious heaps of manuals and find the “diamonds in the rough”. My compliments to the document writers; while not perfect by any means, the amount of low-level details has significantly increased over the years.
Robert Freeman |
Given that my Oracle Press Oracle9i and 10g New Features books have sold very well, I’d say that there is always a great amount of interest in new database releases. Adoption though is another issue.
There are a number of reasons that adoption is slow, of course. The costs of testing applications in new environments can be quite high, as can the risk of upgrading. Of course, as pointed out before, one has to be cautious about new features… early on they do have a tendency to break, and at the worst times. Also, many DBA shops simply do not have the resources to dedicate to upgrades. Finally, I am dismayed at the comment from Joel about his DBA’s. Someone needs to kick them in the pants. It is, in my opinion, simply inexcusable that a professional would not seek to keep current in the tools of his trade. Granted it’s hard to do, but it’s important. Imagine if your doctor didn’t stay current with the more recent developments in medicine. Don’t be lazy out there DBA’s!!
The 11g documentation looks to be quite good in the Beta, so I have hopes that the production set will be outstanding. Oracle support does have some challenges to deal with, and to me this may well be one of the major hurdles to successfully moving the Oracle database community to 11g. Let’s hope that we can get to the point where first level support becomes useful, rather than seen as a rock in the stream (or Dam as the case may be).
Given that most of our apps are third party and from Oracle and none of them makes any use of any of the new features since 9i, it’s very hard to mount a case for upgrading to new versions of the database where new functionality depends on the application code being prepared to use it.
Of courfse, features that can be exploited without dependency from the application are most welcome. Just exactly: which are they?
Releasing 11g seems a bit early..say by 18 months.Why ?
Well i dont see lots of databases on 10g.Almost 70% database to my knowledge are still on the previous versions.
Upgrade issue is decided on only one criterea: ROI.
To learn the new release would take ample time.By learn i mean trully harness its capabilities.
It would have have been better for Oracle to focus entrely on CRS and GRID.And backport these to 10g.That would have been better for Oracle suctomers and eventually Oracle.And we would have seen a flurry of movement to enhanced version.
Now we have the features spread over 2 versions.
10g which many use.
11g which has the best features but none use.
But then cant expect utopia.
The things we use the most in new releases are those that improve the core database system (e.g. the scheduler, datapump, AWR, locally managed tablespaces, dml merge, etc.). There are other great features that are introduced, but just aren’t viable or worth the time, at least for us, but I’m sure many companies do.
Overall though, I am always very happy with what I get from new Oracle releases and we usually move to the Release 2 version when it arrives.
I suspect many of these releases are just stepping stones to the Oracle Appliance. When ASM showed up I wondered how long til you could buy a server and just install Oracle, or just buy an Oracle server. Now with Oracle Enterprise Linux out, I think that might be closer than we think.
Rama Balaji |
To answer the question
What is the process like for rolling out new features in your organization, and how long does that process take? Also, is Oracle helpful on that front?
I read somewhere that “Business matters, technology doesn’t”. This has been a mantra of many companies. As far as I know every DBA who wants to survive in the field, would like to learn New features of every oracle release or atleast curious about them. But DBAs don’t make business decisions, it is made at the upper level who care more about keeping up with their SLAs more so in running their business for more profit or without any loss.
So unless the new features befits or helps them to keep up with their SLAs, they will not implement it. I work at a shop where 80% of their databases still run on 8.1.7. We have a migration plan only to keep the support. We will most likely disable the 10g features and make it run just as 8.1.7. Only when the business need arise, would we even be thinking about using the new features.
However the new projects would be on the new release, RAC enabled etc. This is a situation in many oracle shops.
Around here we tend to upgrade a DB when the Oracle support is one year away of ending. That usually brings us to a already tested and very discussed DB version.
There are several reasons:
– upgrades are painful, both on planning, resources, etc.
– there is always risks involved
– not everyone is eager to jump into a lot of new problems when the current ones are now iron down and DBAs know how to deal with the ones that left.
Bottom line: don’t change a winning team…
Robert G. Freeman |
>> So unless the new features befits or helps them to keep up with
>> their SLAs, they will not implement it.
You have just made my argument for me. Without knowing what feature sets are available, how will you sell the idea that this feature, or that feature, will reduce outages and thus help meet SLA’s? How can you cost justify anything without knowing whats out there?
Any why the concentration on just the work today, when there is a whole tomorrow that is unsure. The word layoff is a very real word, you must be prepared. I can think of bushels of DBA’s who were laid off and found out just how far behind the curve they really were. I’ve lost count of how many I’ve interviewed and turned down for jobs based on very rusty, or limited, skill sets.
If you don’t know which feature sets are available, how can you tell management they are barking up the wrong tree with a given design? How can you put forth the best possible design?
Perhaps management will not listen the first time, or the second time when you tell them they are incorrect. However, when you win yourself as a reputation as the one who was right, time and time again, and you market that carefully (that is, don’t upset people in doing so), there will come a time that they will start to listen to you and you can make a difference.
You should never give up the ship because it has little leaks in it. You should also always have that lifeboat in sight.