April 26, 2011 at 9:22 pm
Comments posted to this topic are about the item Showing Its Age
Follow me on Twitter: http://www.twitter.com/way0utwest
Forum Etiquette: How to post data/code on a forum to get the best help
My Blog: www.voiceofthedba.com
April 27, 2011 at 1:44 am
We have a number of legacy apps and hardware going on 9 yrs. and still doing exactly what they are supposed to.
That said we are in a round of hardware and software updates moving to windows server 2008 r2, sql server 2008 r2, .net, etc... but the "new" hardware will all be pre-owned or re-certified (50 cents or less on the dollar); with some models already 2 to 3 years old. The fact of the matter is; why use a sledge-hammer when a regular hammer works just fine; and a 3 year old server is still going to be quad-core-xeons with a boat load of memory. A one or two year old server will even have DDR3, QPI, SAS Raid, etc....
I have to tell you that the next round of hardware/software is going to have to do something pretty spectacular, or if the hardware begins to fail, to warrant the next update. Hopefully we can get another 8 years!!!!
April 27, 2011 at 5:59 am
Good editorial... But lets be honest about it, each year or so companies and individuals all over the world "upgrade" their software for no good reason at all. SQL Server aside - which does show improvements and enhancements with successive versions - most Microsoft products are not "new products" at all. In fact, a recent study showed that the "upgrades" to the last 5 versions of Microsoft Office had LESS than 2% of 'upgrades or enhancements'.
But we have been trained much like Pavlov's dog. The bell rings from Redmond and we rush out to buy the latest repeat of the same thing we already own. As one analyst put it; "If you are simply writing letters to your friends with Word, wouldnt Word 95 accomplish the exact same task as Word 2010?" yes, it would.
Hardware of course, is a bit different and yes, we do see 'improvements' therein, however if you look closely at certain business units - say manufacturing - they still run a number of DOS based systems on 20 and sometimes 30 year old dependable hardware! Why? Because they work, and they work without all the bells and whistles that really, serve no purpose at all.
The DMV in Colorado might have been better able to handle their mess if we took a longer view of support for both hardware and software. But we dont. Why? Because doing so would mean companies in the business would not be making the millions they are. In other words, its all largely marketing and sales that matters.
There are some things in life that cannot be improved. You can strap a jet engine to a hammer, maybe add an iPod and cell phone to it, and allow it to download apps. But ask yourself; when you need to bang a nail in, the hammer that worked well 100 years ago is still the best solution to this day - and any 'enhancements' just get in the way. Maybe we should learn from that, but then sales and marketing would not have much reason to try to baffle us with BS...
April 27, 2011 at 7:16 am
A situation I have run into a is old software that is no longer supported by the vendor or the vendor is out of business. Getting old software to run on a newer OS can be a real problem, or you may not have the install media. When that 10 year old hardware running an important but unsupported application starts to fail, what do you do?
For these cases, doing a physical to virtual migration can be a good solution, because it gets the ancient hardware out of the picture. Just bring the whole system over intact to a VM, and keep going on new hardware. It doesn’t solve every problem, but if that app running on Windows NT 4.0 and SQL Server 7.0 is working OK, why upgrade?
April 27, 2011 at 7:39 am
Michael Valentine Jones (4/27/2011)
A situation I have run into a is old software that is no longer supported by the vendor or the vendor is out of business. Getting old software to run on a newer OS can be a real problem, or you may not have the install media. When that 10 year old hardware running an important but unsupported application starts to fail, what do you do?For these cases, doing a physical to virtual migration can be a good solution, because it gets the ancient hardware out of the picture. Just bring the whole system over intact to a VM, and keep going on new hardware. It doesn’t solve every problem, but if that app running on Windows NT 4.0 and SQL Server 7.0 is working OK, why upgrade?
I had a friend do this with a keycard system. Worked fine, but vendor went out of business and it ran only on Nt4.0. Went P->V, running fine on a 1 socket, 1 core, 2GB RAM system. Probably live like that forever since the upgrade to a newer keycard system is about 100K with no new features.
Follow me on Twitter: http://www.twitter.com/way0utwest
Forum Etiquette: How to post data/code on a forum to get the best help
My Blog: www.voiceofthedba.com
April 27, 2011 at 7:43 am
blandry (4/27/2011)
Good editorial... But lets be honest about it, each year or so companies and individuals all over the world "upgrade" their software for no good reason at all. SQL Server aside - which does show improvements and enhancements with successive versions - most Microsoft products are not "new products" at all. In fact, a recent study showed that the "upgrades" to the last 5 versions of Microsoft Office had LESS than 2% of 'upgrades or enhancements'.But we have been trained much like Pavlov's dog. The bell rings from Redmond and we rush out to buy the latest repeat of the same thing we already own. As one analyst put it; "If you are simply writing letters to your friends with Word, wouldnt Word 95 accomplish the exact same task as Word 2010?" yes, it would.
I think some of that is changing. My wife has a 4 year old laptop from her company, running XP and Office 2003. She is getting an upgrade this summer, b/c of hardware. She'll move to Win 7 and Office 2010 just because those are the current versions. She'd like to just move to Office 2007 since we have that at home and it works, but the "trick" of making PPTs in 2010 not backwards readable has stuck her into the Office 2010 upgrade.
This is one place I'd like to see some govt standards. Make programs like PPT be able to save a particular format as a default. Then we wouldn't see customers and partners forcing us to upgrade when they do.
Follow me on Twitter: http://www.twitter.com/way0utwest
Forum Etiquette: How to post data/code on a forum to get the best help
My Blog: www.voiceofthedba.com
April 27, 2011 at 7:56 am
Steve Jones - SSC Editor (4/27/2011)
This is one place I'd like to see some govt standards. Make programs like PPT be able to save a particular format as a default. Then we wouldn't see customers and partners forcing us to upgrade when they do.
I don't know if the best way to handle this is with government standards but I completely understand what you're getting at. I do have to agree that without regulation or immense pressure this isn't likely to change though.
April 27, 2011 at 8:17 am
I hear you, and overall I dislike government regulation as it's been done. However companies, and capitalism fails because of overwhelming greed and a race to the bottom. People start to do anything to make a $$, or more a couple more $$ than their friend instead of reaching a sustainable level.
More I think government provides a framework and low bar to be met that ensures safety, security, privacy, and competition and then companies move beyond that. Change your format, fine, but I have to have a default in Word 2012 or Wordperfect: Zombie edition that allows me to select the 2003 format and it remains there by default. It also doesn't let me put in features in my doc that don't work backwards
Follow me on Twitter: http://www.twitter.com/way0utwest
Forum Etiquette: How to post data/code on a forum to get the best help
My Blog: www.voiceofthedba.com
April 27, 2011 at 8:30 am
The last company I worked for is still using SQL 2000 with an ERP application that hasn't been supported by the vendor in at least 5 years. I wrote an application partly as a proof of concept and partly to solve a short term problem that I expected would be replaced within a year or 2. That application lived almost 8 years.
My point is that this is reality, and we need to start thinking, developing, architechting with much longer time horizons than we might prefer. Think of the longest horizon you need to work with, and then add at least 10 years to your thinking. Do you think mainframe software folks in the 1970's were worried about the Y2K issue when they worked-NO-they didn't believe their applications would be around that long. Big mistake.
April 27, 2011 at 9:11 am
The cost of upgrading versus the actual benefits of doing so seems to be the main issue that keeps legacy systems around. The vendor support problems are bad enough on older systems but the compatability issues are what really gets me. There comes a point when it is so hard to integrate a legacy system into more modern systems that this alone should justify the cost of upgrading. The legacy system becomes an island that you can't do anything with and a roadblock for future system integrations. By the time that you build custom applications to make the old apps work with the new ones, you could have used those labor hours towards a system upgrade. You would probably end up with a new support contract as well and a more capable system with some ability to grow.
Sadly, justifying these sorts of upgrades is usually a hard sell unless a major outage occurs and the age issues around the legacy equipment shows.:(
April 27, 2011 at 9:22 am
Daniel Bowlin (4/27/2011)
The last company I worked for is still using SQL 2000 with an ERP application that hasn't been supported by the vendor in at least 5 years. I wrote an application partly as a proof of concept and partly to solve a short term problem that I expected would be replaced within a year or 2. That application lived almost 8 years.My point is that this is reality, and we need to start thinking, developing, architechting with much longer time horizons than we might prefer. Think of the longest horizon you need to work with, and then add at least 10 years to your thinking. Do you think mainframe software folks in the 1970's were worried about the Y2K issue when they worked-NO-they didn't believe their applications would be around that long. Big mistake.
A company I worked for had a database (non-relational) that was implemeted in 1971 with a single digit year field because the designer didn't think it would be in use in 10 years.
In 1980, they added an additional 1 digit field for the decade. :unsure:
April 27, 2011 at 12:04 pm
About half of our stuff is still in SQL 2000. One large SQL 2000 Cluster will be migrated to SQL 2008R2 this Summer due to the app upgrade. However, last summer we implemented a new store polling system that the software company still did not support Win2008 or SQL 2008 so we had to install it on Win2003/SQL2005 two clusters and two standalone SQL Servers... crap!
The amount of money to replace all of the underlying hardware and SQL Server license upgrades is big $$ for very little bang for the buck in all reality. I inherited four very complicated old hardware SQL2000 installed in 2001 that has gone through four of five developers/dbas and everyone is afraid to touch them.
April 27, 2011 at 1:47 pm
Nice photo in your editorial, Steve.
I remember when tape drives like the one in your photo were state of the art. Disk drives had just begun to make their appearance in the mini-computer realm.
LC
April 27, 2011 at 6:00 pm
There really three stories to this: hardware, software and business. Each has a completely different life cycle, but are often tied to each other at the time they were developed.
- Hardware shows wear and tear, so even if existing hardware is fully up to the task at hand, after a while it will start to fail and needs replacing.
- Software itself does not degrade over time. An algorithm or data-structure fully up to the task at hand will remain so indefinitely. But time may show vulnerabilities which require patching or accumulate more data over time that could cause capacity issues requiring adding bigger hardware or adding archiving functionality not thought of in initial development.
- Businesses need to deal with changing markets, own growth and will require to change their methods accordingly making changes to the tasks at hand, rendering the software and/or hardware designed for the old task no longer suitable.
The reality is that all three parts interact with each other and changes in requirements on one part will affect suitability/compatibility with the other parts.
Businesses have been around for a very long time and there is a lot of experience in dealing with change and redefining the tasks to make the business work, improve and grow.
But both hardware and software designers are only now gaining more and more experience in designing for change. Technologies and methodologies have been emerging in the past decade to facilitate change while at the same time ensuring compatibility and interoperability. Even advances in metallurgy makes a modern hammer better than a hammer from 100 years ago (higher resistance to wear and tear for example).
The biggest difference between computer science and other (older) technologies is the pace in which advances are made over time, which is posing new challenges to the processes to deal with change.
We'll learn, but in the meantime we will have to deal with and accept the occasional outages caused by the mistakes made in the past due to inexperience in this field.
Marco
"The only constant is change".
April 28, 2011 at 4:47 pm
I notice everyone always cites greed as the reason that manufacturers come out with new version s and stop supporting old versions. But at that same time we can't have a rose colored glasses view of support either.
You don't expect your auto manufacturer to support your car forever at the fixed cost of your initial purchase. You've grown used to a scale of costs from paying for oil changes through replacing a major assembly or turning over your vehicle for a new one.
In our discipline if we were willing to accept a service model for some of our products where we pay a yearly fee for support (and some DBMS's do have that model) then the manufacturer will gladly support that system for as long as you want to pay the fee. Sure, they may offer you enticements to lower their own support costs... Say free upgrades to the new software version so they don't have to train their techs in as many versions. But, overall, if there's money there to be made they'll keep supplying the service.
But the bottom line is when you pay some amount up front for your system that price has to be stretched out to cover a laundry list of things from manufacturing costs to development costs to support costs. If over time the costs keep adding up to the manufacturer and the money has long since been spent then sooner or later they're going to have to stop spending money on that old system.
It's fine to talk about hammer and how they did the same job they did a thousand years ago but that's not the same kind of situation that a DBMS manufacturer faces. You want ongoing security patches? Then "There ain't no such thing as a free lunch". It's the consumer that has lead the industry down the path to constant rebranding and changing the face without big feature set changes. We are willing to pay for newer shinier stuff (even if it's just gilding the lily) but not one penny for fixes because "I already paid for that once" then the market evolves to fit our demands.
If we want to talk about building a smarter market and making software investments last longer then, I think, we need to be willing to look at the whole model and take a realistic look at the costs. If we are willing to buy software as if it were an ongoing relationship instead of a single purchase commodity then manufacturers will be willing to sell it that way.
Viewing 15 posts - 1 through 15 (of 19 total)
You must be logged in to reply to this topic. Login to reply