October 3, 2008 at 8:42 am
Charles Kincaid (10/3/2008)
Then, more on topic, I have customers that run their entire business on SQL Server Express. They already had the machine, OS, and AV. We show them how to back it up. We even worked with them to develop a cmd file to make the process easy. We used our tiny scheduler to automate the process. (Can't run scheduled jobs in Express. 😎 Command line interface to the rescue.) They had purchased a few days of on-site configuration an support services anyway. What is their cost for SQL. Say it with me friends. ZERO. The performance is good enough for them. When they do outgrow that then we upgrade to the full version and license that per server/processor.
Is it fit for purpose? Is it effective? Is it cost-effective?
Yes, you say? My God, those businesses are in dire danger of staying solvent if they make more practical decisions like that.....
Semper in excretia, suus solum profundum variat
October 3, 2008 at 8:56 am
I think everyone agrees that the price/performance ration is the key driver for business, but there's usually also a minimally acceptable performance to be taken into account.
Sometimes, any performance is acceptable so minimum cost (including hardware etc) wins, but usually there's some minimum needed, which is where the absolute performance has to be considered.
Sadly, I've never been in an unlimited budget situation where the only factor was hitting performance. However, that's where it gets interesting as you have to look at the design, code, process, etc and try to tweak a bit extra out of the best hardware/software the budget will allow. 🙂
Derek
October 3, 2008 at 9:49 am
I disagree with your comparison of Texas and Texas A&M. Its actually the opposite, A&M has spent almost as much in many areas of athletics, primarily facilities and coach's salaries however they have gotten little in return.
Texas has 47 national championships (http://texassports.com/trads/tex-championships.html) across a variety of sports, while the only 1 I know of for A&M was back in 1939 for football.
Hook 'em!
[font="Arial Narrow"]bc[/font]
October 3, 2008 at 10:07 am
What appears to be the norm that I hear of is that an organization buys the fastest they can afford and prays that it is fast enough.
What I think should be the case is determine what you think you need, add growth, add the unforseen factor, and then buy 35 more then that total.
We have to remember that we require enough machine, installed application, and service to make the solution work with the required speed. But in a growing business where you are expanding your business and service areas, you buying of IT equipment is in some ways like buying clothes for kids. If you buy what fits today before it has been used three times it is too small.
Buy once large or many small, will keep the price down some. But always buy the speed you need. If you can not do it and do it fast enough to keep the user happy, you can and will be replaced.
Miles...
Not all gray hairs are Dinosaurs!
October 3, 2008 at 10:25 am
I vote for price/performance but with a eye on growth potential. Much easier to add drives and memory to extend the life and speed performance than starting out with a fully loaded and under used server that allows for sloppy use such as large transaction files to be created. When you realize you are in trouble, you have no place to go except new.
October 3, 2008 at 12:57 pm
Long term, the gold standard is cost of ownership for many companies, especially enterprises that are developing a Tier 1 utility database platform. For a database, that includes DBAs, training, upgrade, servicing (if you have to pay for service packs on the Oracle altar) and access to documentation/knowledgebase (if you have to pay for access to the Oracle altar)... and the list goes on...
There are situations where TCO isn't the crux, though, because "money is no object." In those situations you bet on what the people you trust know and what you think you can deliver on in the time available... and that's probably why most vendors don't spend time or money on benchmarks, because they've reached the point where it's about relationship selling for them and brand, not price/performance ratio or some other objective measure.
October 3, 2008 at 2:13 pm
The price/performance issue is really part of being a good crafsperson ( PC huh!). You write the best code you can within the limits of the dead line imparatives. You don't RBAR. You do use set updates. Program comments are a requirement.
If you don't persue good craftsmanship then you and your client lose.
October 3, 2008 at 2:29 pm
As someone else mentioned, when it comes to cost, I have to add in reliability as a major factor, not just speed.
RAID-0 is cheap and high performance. I'd never put a production database on it unless there was some other means of recovering from a lost disk (server mirroring, maybe; but that increases the cost).
Mirrored servers aren't cheap, and don't improve performance, but something for continuity is in order for most businesses and databases. Log shipping might be slightly cheaper, but still it's money spent on zero net increase in performance.
As far as just the question of performance or performance:cost ratio, I'd have to go with the ratio. Unless we're talking about a gaming machine, that's just the way I'd have to look at it. Even if you have extra budget, if you can get a machine that will suite forseeable needs with some margin for error, for less than the full budget, then spend the rest on redundancy or something else.
As already mentioned, even a "slow" server can usually be sped up with better code, better tables, better indexes, a little maintenance (defrag indexes, etc.), and so on.
- 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
October 3, 2008 at 3:15 pm
Nobody has unlimited budgets, so price is always a limiting factor. Therefore I cast my vote on the price/performance side of the house. Not that he needs it, but I agree with Mr. Moden. Good design and good code means less of a load on the servers, which means that your existing hardware can handle higher volumes (through growth or new applications) without needing as many upgrades. The longer you can stretch your existing equipment's life, the better price performance you will get when you are finally forced to replace or upgrade. If nothing else, it also means that your queries and procedures deliver results fast, and as a wise man once told me: "Anything longer than instantaneous is too damn slow."
With that said about hardware and code, how does anyone else feel about the whole issue of performance vs price/performance as applied to people? Sometimes you want to fire three weak coders to free up the money to get one really good one. Sometimes you want enough warm bodies to handle the immediate-but-trivial jobs that keep your top people from focusing on the longterm-but-critical work. I'm still undecided about this.
__________________________________________________
Against stupidity the gods themselves contend in vain. -- Friedrich Schiller
Stop, children, what's that sound? Everybody look what's going down. -- Stephen Stills
October 3, 2008 at 5:58 pm
Price/Performance? Is that an old-fashioned concept or what?
Haven't we all learned yet that price doesn't matter anymore? The news of the past few weeks shows that you should just buy the biggest, baddest (fill-in-the-blank) you want, regardless of your ability to pay. Then when the bill comes due, get together with everyone else who did the same thing and go to Washington asking for a bailout.
Problem solved! 😎
October 3, 2008 at 6:05 pm
Ruh-roh. Bob's pulled in politics... a Godwin event can't be far behind. :hehe:
October 4, 2008 at 4:09 am
A very interesting topic. We want quality/smart code which will enhance the performance of a slow system as said by Jeff. But to write a smart code you have to learn and practice lot. 😛
October 4, 2008 at 4:13 am
A very interesting topic. We want quality/smart code which will enhance the performance of a slow system as said by Jeff. But to write a smart code you have to learn and practice lot. 😛
October 4, 2008 at 11:31 am
Heh... no... writing smart code isn't difficult. You simply have to make the choice not to write bad code. 😛
--Jeff Moden
Change is inevitable... Change for the better is not.
October 4, 2008 at 2:18 pm
Did you wrote smart code in Day 1? I believe practice, knowledge and experience makes one smart developer. It may be smart developer but not genius like you. 😉 Genius are rare.... 😛
Viewing 15 posts - 16 through 30 (of 32 total)
You must be logged in to reply to this topic. Login to reply