December 26, 2009 at 12:10 pm
Comments posted to this topic are about the item Service Pack 4
December 26, 2009 at 8:23 pm
We are a 60/40 SQL 2005/SQL 2000 shop - yes I agree that testing each and every CU is a pain and not possible with over 330+ instances - we go with a new CU anywhere from 3-6 months (unless there is a compelling need). Right now on SQL 2005 we are either at CU 3 or CU 4. As for 2000, well SP4 and a few hotfixes depending on the application
However SQL 2005 SP4 is an absolute must in my opinion.
The primary reason is that we have no intention of looking at SQL 2008. As for SQL 2008 R2, well not at least until until at least SP2 !
Who knows, by that maybe we'll be at SQL 2012 and can play 'the game' again !!!
RegardsRudy KomacsarSenior Database Administrator"Ave Caesar! - Morituri te salutamus."
December 27, 2009 at 3:11 am
I guess I don't understand the supposed dilema here... if Cumulative Updates really are cumulative and include all fixes since the last service pack, isn't that just as good as a service pack? Why not just pick one CU to test and settle for that?
As a sidebar... now you know why some customers just don't like "agile" development methods. The updates are a PITA. 😉
--Jeff Moden
Change is inevitable... Change for the better is not.
December 27, 2009 at 12:12 pm
Jeff Moden (12/27/2009)
I guess I don't understand the supposed dilema here... if Cumulative Updates really are cumulative and include all fixes since the last service pack, isn't that just as good as a service pack? Why not just pick one CU to test and settle for that?
MS themselves say they are not as good as service packs, if MS don't guarantee the results of applying CU's, personally I am steering clear of them. Go to the latest SP + any security patches and stick there
Remember how long we had to wait for SQL 2005 SP3 and the campaigning that appeared to be necessary before it was released? It looked at the time as if all the effort went into SQL 2008 release,and perhaps SP4 is suffering a similar fate. SP3 seems pretty stable so I am in no hurry for an SP4, when the backout of a service pack is an uninstall\reinstall who needs the stress?
An SP2 for SQL 2008 would be nice at this point. rather than muddying the waters even more with a release 2
There was never a final service pack for SQL 2000, just a hotfix rollup MS recommended you applied (as opposed to CU's which they don't recommend you apply). Even then there have been post SQL 2000 SP4 security patches (MS08-040) that have had to be applied.
---------------------------------------------------------------------
December 27, 2009 at 1:15 pm
The CUs aren't regression tested, and do sometimes introduce issues. SPs include the CUs, plus potential internal fixes for things that were in a customer specific hot fix, or might fix another issue that was noticed. they are also regression tested with the full test suites.
Personally I like the CUs in terms of addressing issues that someone might have, but for keeping patched on the majority of servers, I'd stick with SPs. One a year seemed like too few, but I'd be happy if we could just get that one.
December 27, 2009 at 4:02 pm
I think of SQL Server/Oracle the way I think of BMW/Mercedes et al. I am paying a premium price for something that comes with certain quality expectations.
Part of those expectations relates to customer service. Quality customer service may be a "cost" to the supplier but is the thing that makes me buy in the first place.
If Product 'x' is only marginally better than Product 'y' and obviously more expensive then it is the periphery and sometimes intangible benefits that make Product 'x' the compelling choice.
I've said it before that costs are easy to measure because they are upfront and clearly visible on a balance sheet. The revenue or benefit is slower to be apparent and may even be hard to measure.
In many ways measuring the benefits is like piloting a hot air balloon. An inexperienced pilot will have them going up and down all over the place because you have to look forward. By the time you start to drop you will continue to drop for some time after you apply corrective measures due to inertia. Similarly you can soar on inertia without realizing that you have very little to sustain you.
December 27, 2009 at 10:20 pm
Thanks for the feedback George and Steve. Didn't know that about CU's. Guess I just got lucky by being lazy... if it's not a service pack or a very specific patch (like security update), it's not going in on my watch.
--Jeff Moden
Change is inevitable... Change for the better is not.
December 27, 2009 at 11:11 pm
I'll go one further - I don't think it's acceptable to have service packs. It's an acknowledgement that a product is defective. Only after an unspecified number of updates (and years) will it work correctly. The worst part is that we accept and, in fact, expect it. It's irritating that MS isn't giving us SP4 this year. But it's even worse that we should even have to expect one.
James Stover, McDBA
December 28, 2009 at 4:00 am
James Stover (12/27/2009)
I'll go one further - I don't think it's acceptable to have service packs.
Brave statement. I'm pretty rigorous when writing code but I doubt that even John Carmack could claim to write bug free code.
If you look down the bug list for SQL Server you will see that the vast majority is the sort of stuff that crops up only in specific circumstances. On a large product such as SQL Server, or any other DB system for that matter, the sheer number of possible circumstances make it nigh on impossible to test all possibilities.
Any complex product has bug fixes and, scarily, that includes aeroplanes. See http://www.aaib.gov.uk/publications/index.cfm
When making changes software is just as subject to the law of unintended consequences as anything else. Little systems that function fine in isolation produce behaviours that are unintended and even undesirable when combined.
I saw an engineering show that mentioned that shutting one of 3 tunnels in a busy intersection actually improved the traffic flow. The expected consequence was that it would make rush hour intollerable! It would be really easy to come across this situation in software with a multi-threaded environment.
December 28, 2009 at 6:21 am
I'm not talking about bug-free code. Practically speaking, that's impossible. I'm talking specifically about the expectation that a software product will require a service pack (or packs) to function correctly. We have been trained by the industry to expect them and we just accept it as standard practice. The software industry should not be exempt from being held to a higher standard just because its products are complex.
James Stover, McDBA
December 28, 2009 at 7:59 am
James Stover (12/27/2009)
I'll go one further - I don't think it's acceptable to have service packs. It's an acknowledgement that a product is defective. Only after an unspecified number of updates (and years) will it work correctly. The worst part is that we accept and, in fact, expect it. It's irritating that MS isn't giving us SP4 this year. But it's even worse that we should even have to expect one.
So, you don't change the oil in your car? You don't brush your teeth? You don't clean the gutters on your house (or have someone do it for you)?
We live in a universe where the laws of thermodynamics apply. That means complex systems decay and require an input of energy to maintain. That means software will require updates. Until you live in a universe with different laws of physics, that's going to be true.
- Gus "GSquared", RSVP, OODA, MAP, NMVP, FAQ, SAT, SQL, DNA, RNA, UOI, IOU, AM, PM, AD, BC, BCE, USA, UN, CF, ROFL, LOL, ETC
Property of The Thread
"Nobody knows the age of the human race, but everyone agrees it's old enough to know better." - Anon
December 28, 2009 at 7:59 am
I voted for both of them. Good thought there, Steve. SP3 for 2005 took that kind of nudge to get going, better to act now on this one than wait as long.
- Gus "GSquared", RSVP, OODA, MAP, NMVP, FAQ, SAT, SQL, DNA, RNA, UOI, IOU, AM, PM, AD, BC, BCE, USA, UN, CF, ROFL, LOL, ETC
Property of The Thread
"Nobody knows the age of the human race, but everyone agrees it's old enough to know better." - Anon
December 28, 2009 at 9:04 am
Thanks for the votes. I've pinged MS, and been patient waiting for the one year anniversary of SP3 to roll by, but no responses. My guess is most of MS is on holiday this week, and that's OK. However we do need the patch to come.
Great analogy on the car. Software does require maintenance and we should expect it. What we should not expect is fundamental errors being left in the product when it's released. I don't think we get that, though SQL SS2K5 had the issue of mirroring not working in RTM and enabled in SP1.
December 28, 2009 at 10:29 am
James Stover (12/27/2009)
I'll go one further - I don't think it's acceptable to have service packs. It's an acknowledgement that a product is defective. Only after an unspecified number of updates (and years) will it work correctly. The worst part is that we accept and, in fact, expect it. It's irritating that MS isn't giving us SP4 this year. But it's even worse that we should even have to expect one.
Just to throw my 2 cents in, there is no way MS can test all the environments that SQL Server will be put on and to what uses it will be used. As they say, "Nothing is fool proof because fools are so ingenious." As a result, no matter how well they code, someone is going to find some situation that needs a fix.
I play online MMO games. The companies that put them out are always doing patches and improvements. Some of these things shake out during a beta test, most of them don't. One game in particular, years ago, I remember as doing patches of patches of patches before they got it to run properly (and this was after the code apparently went thorugh QA to be released to the client).
The game I currently play patches about every month or so, sometimes more, sometimes less. Not all repercussions of minor code changes are found or caught. I don't mind the patches as each one pretty much improves the overall game. Same thing applies to other software products.
Things change, it's how life is. If the company couldn't change their software to adjust to customer wants and needs, then it wouldn't be worth much in the end, now would it?
-- Kit
December 28, 2009 at 3:18 pm
Besides bug fixes, service packs can contain functional and performance improvements, as well as completely new functions. I look forward to them...but I also don't have to apply them. So we have a choice, and I like that.
LinkedIn - http://www.linkedin.com/in/carlosbossy
Blog - http://www.carlosbossy.com
Follow me - @carlosbossy
Viewing 15 posts - 1 through 15 (of 24 total)
You must be logged in to reply to this topic. Login to reply