Hi Guys,
I have a client who hopes to keep their server another 5 years making it 10 years old at that time.
At this point there are no plans to add new infrastructure or a new server to the mix. Their business model is fairly static.
I'd like to see them upgrade. Can anyone suggest specific reasons why running a business on 10 year old equipment is a bad thing?
Specific arguments I can think of would be: - Hard/Impossible to find replacement hardware - Lack of support for both H/W and S/W - Possibly unable to run current versions of CentOS - Higher probability of hardware failures over time - Performance bottlenecks
Any other thoughts?
Shawn
Good one.
I run into very similar situations.
Focus purely on cash cost of maintenance of older stuff vs newer stuff.
Quote reputable sources like Gartner Group, etc...
Get a little familiar with ITIL in terms of like cycle.
Its very daunting to convince companies to spend money but if you frame your Power Point, Project presentations around road A costs this much, road B costs that much, you pick, then mebbe good things will happen.
Another thing my mom has instilled is that "you catch more bees with honey".
I have a tendency to sound like I am preaching and I am not, in fact my motto is "I dunno shizzle".
But I just want to emphasize the pain that I run into this all the time and some times succeed, I pretend to be a CFO/bean counter with the attitude "if it ain't broke, why replace it", which is valid if you think of it.
On Oct 9, 2009, at 3:29 PM, Shawn Everett wrote:
Hi Guys,
I have a client who hopes to keep their server another 5 years making it 10 years old at that time.
At this point there are no plans to add new infrastructure or a new server to the mix. Their business model is fairly static.
I'd like to see them upgrade. Can anyone suggest specific reasons why running a business on 10 year old equipment is a bad thing?
Specific arguments I can think of would be:
- Hard/Impossible to find replacement hardware
- Lack of support for both H/W and S/W
- Possibly unable to run current versions of CentOS
- Higher probability of hardware failures over time
- Performance bottlenecks
Any other thoughts?
Shawn
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On Fri, Oct 09, 2009, Shawn Everett wrote:
Hi Guys,
I have a client who hopes to keep their server another 5 years making it 10 years old at that time.
At this point there are no plans to add new infrastructure or a new server to the mix. Their business model is fairly static.
I'd like to see them upgrade. Can anyone suggest specific reasons why running a business on 10 year old equipment is a bad thing?
Specific arguments I can think of would be:
- Hard/Impossible to find replacement hardware
- Lack of support for both H/W and S/W
- Possibly unable to run current versions of CentOS
- Higher probability of hardware failures over time
- Performance bottlenecks
Any other thoughts?
These are plausible reasons, but may not be in the client's best interests, assuming that you are monitoring the system for potential hardware problems and have good, verified backups that can be used to build a new system quickly if necessary.
We have many commercial clients doing accounting/database things where the existing hardware and software does everything they need and want, and their systems are behind good firewalls so they're not generally exposed to the Internet (and none run Windows). They much prefer to put off hardware replacement until there's some good or compelling reason to do so (e.g. one of our clients is running SCO OpenServer 5.0.6a on hardware installed in November 1999, and we're planning on building a CentOS box with VMware to replace it after first of the year).
Most of our clients are small-to-medium businesses that only work one shift, not huge enterprises going 24/7 where zero down time is a must. Uptimes on these systems is often measured in years, and outages are mostly due to an Internet connection going down when the telco screws up. Hardware updates are usually done by building a new server in parallel with the production box, then rsync'ing data over at close-of-business on Friday night with the weekend available to deal with visits from Mr. Murphy.
Bill
On Fri, Oct 9, 2009 at 6:29 PM, Shawn Everett shawn@tandac.com wrote:
Hi Guys,
I have a client who hopes to keep their server another 5 years making it 10 years old at that time.
At this point there are no plans to add new infrastructure or a new server to the mix. Their business model is fairly static.
I'd like to see them upgrade. Can anyone suggest specific reasons why running a business on 10 year old equipment is a bad thing?
Specific arguments I can think of would be:
- Hard/Impossible to find replacement hardware
- Lack of support for both H/W and S/W
- Possibly unable to run current versions of CentOS
- Higher probability of hardware failures over time
- Performance bottlenecks
So they're running a five year old version of centos now? There's no reason a server if properly maintained shouldn't run for ten years. Upgrading just for the sake of upgrading isn't much of a justification. Unless the machine is showing signs of degradation, there's no reason to fault the hardware. Software support is a subjective problem; software support for what exactly? If the machine has made it five years with no hardware failures, anything that was going to fail should have by now. Slow performance might be a valid point, but it ties into the software question, what exactly does the server do, is high performance a big issue?
One of my customers is running a 12+ year old SCO machine right now, and it's running custom software created more like 20 years ago. The vendor that set it all up is a "one man" operation, and it looks like he could fall over at any minute. The data is all locked up in thousands of files that comprise some sort of crazy proprietary database. The server's power supply has been complaining lately, so the client went to the vendor for a quote on fixing the problem. The vendor's solution is to spend thousands and thousands for new hardware and to "upgrade" the software. Researching the problem I contacted a regulatory agency that overseas the operation of my client and found documentation it was strongly advised ten years ago that they stop using the vendors software and hardware. That advice was ignored and now they have an expensive mess with no clean way out.
Lots of fun! -Gordon
On Fri, Oct 09, 2009, Gordon McLellan wrote:
On Fri, Oct 9, 2009 at 6:29 PM, Shawn Everett shawn@tandac.com wrote:
Hi Guys,
I have a client who hopes to keep their server another 5 years making it 10 years old at that time.
At this point there are no plans to add new infrastructure or a new server to the mix. Their business model is fairly static.
I'd like to see them upgrade. Can anyone suggest specific reasons why running a business on 10 year old equipment is a bad thing?
Specific arguments I can think of would be:
- Hard/Impossible to find replacement hardware
- Lack of support for both H/W and S/W
- Possibly unable to run current versions of CentOS
- Higher probability of hardware failures over time
- Performance bottlenecks
So they're running a five year old version of centos now? There's no reason a server if properly maintained shouldn't run for ten years. Upgrading just for the sake of upgrading isn't much of a justification. Unless the machine is showing signs of degradation, there's no reason to fault the hardware. Software support is a subjective problem; software support for what exactly? If the machine has made it five years with no hardware failures, anything that was going to fail should have by now. Slow performance might be a valid point, but it ties into the software question, what exactly does the server do, is high performance a big issue?
Agreed. Quality hardware can last a long iime. We have a client with an SCO OpenServer 5.0.5 system still in production that we installed in December 1999. They have been planning on replacing it for years, but it just keeps on going.
One of my customers is running a 12+ year old SCO machine right now, and it's running custom software created more like 20 years ago. The vendor that set it all up is a "one man" operation, and it looks like he could fall over at any minute. The data is all locked up in thousands of files that comprise some sort of crazy proprietary database. The server's power supply has been complaining lately, so the client went to the vendor for a quote on fixing the problem. The vendor's solution is to spend thousands and thousands for new hardware and to "upgrade" the software. Researching the problem I contacted a regulatory agency that overseas the operation of my client and found documentation it was strongly advised ten years ago that they stop using the vendors software and hardware. That advice was ignored and now they have an expensive mess with no clean way out.
FWIW, that 12+ year old SCO system can probably run quite nicely under a VMware virtual machine, and be significantly faster than it is today. You won't get support for hardware like Specialix multi-port boards, so may have to replace a bunch of dumb serial terminals with networked devices, but that's certainly less than tens of thousands of dollars to ``upgrade'' (where I have often seen vendor's ``upgrades'' from SCO to be much worse to use than what they replace).
As much as I despise SCO's actions since Darl and company took over, their systems tend to run without problems for many years. The expense to move off the SCO systems can be prohibitive for a small business, so keeping them going on a CentOS box running VMware or other virtualization software can be a very reasonable solution.
My job is to provide the best service for my clients, which may well mean keeping old stuff running as long as possible.
Bill
On Fri, Oct 9, 2009 at 7:28 PM, Bill Campbell centos@celestial.com wrote:
FWIW, that 12+ year old SCO system can probably run quite nicely under a VMware virtual machine, and be significantly faster than it is today. You won't get support for hardware like Specialix multi-port boards, so may have to replace a bunch of dumb serial terminals with networked devices, but that's certainly less than tens of thousands of dollars to ``upgrade'' (where I have often seen vendor's ``upgrades'' from SCO to be much worse to use than what they replace).
\ gripe mode That was my recommendation originally, either capture the existing install, or perform a new install in a vmware/citrix VM. However, the vendor chose to do their programming in IMS Basic with the license tied to undisclosed hardware signatures of the existing machine. Additionally IMS supposedly will not license their software to run in a virtual machine, according to the vendor who wants us to buy a new server from him. The best part the vendors deal, the "upgrades" will cost us the same price if we buy the server from him or not - opting not to buy overpriced hardware causes the programming fee to inflate by an equal amount - how is that for service! \ gripe mode off
-Gordon
2009/10/10 Gordon McLellan gordonthree@gmail.com:
The best part the vendors deal, the "upgrades" will cost us the same price if we buy the server from him or not - opting not to buy overpriced hardware causes the programming fee to inflate by an equal amount - how is that for service!
Sounds like a fantastic business model though!
However, could be cheaper in the long run for them to look for an alternative solution now.
Take care,
Ben
I'm in the position of having both the mission critical & regular servers.
Our mission critical machine, an app server that runs about 90% of our business, we lease so it gets recycled every five years. For us, it's worth the extra cost to have equipment under warranty with parts guaranteed next business day and a local IBM rep who can be on-site within an hour if something goes wrong.
The less critical machines we actually plan to push out to ten years if possible. Those machines, also IBM's, run our email, website, file serving, and some small web apps we've developed in house. We can tolerate downtime on those machines for a day or so while I scrounge parts. And given we don't expect to be outgrowing those machines any time quick, I can't justify the cost of new servers.
On Fri, Oct 9, 2009 at 4:29 PM, Shawn Everett shawn@tandac.com wrote:
Hi Guys,
I have a client who hopes to keep their server another 5 years making it 10 years old at that time.
At this point there are no plans to add new infrastructure or a new server to the mix. Their business model is fairly static.
I'd like to see them upgrade. Can anyone suggest specific reasons why running a business on 10 year old equipment is a bad thing?
To be honest, I don't think any of the above are reasons to upgrade for many businesses.
Specific arguments I can think of would be:
- Hard/Impossible to find replacement hardware
There are tons of 486 systems still at various surplus places. Getting replacement hardware is going to be harder but you need to show actual numbers of availability and costs and time to replace. In some cases, the business will just decide to buy 4 boxes to scavenge from and call it good. And for a lot of places that makes perfect sense.
- Lack of support for both H/W and S/W
Most likely the hardware is already out of support. Depending on which version of S/W they are running it is less likely to be out of support. If they are using CentOS-3 then they are going to be Out of Support next year. But to be honest for a lot of places they aren't set up to upgrade to anything newer until it becomes a crisis for them.
- Possibly unable to run current versions of CentOS
This could be a catch 22 for them. The applications they may have may only run on CentOS-3 or 2 and aren't something they can afford to update to something newer. [Either the software is no longer updated or the cost model for an update is prohibitive to them.] So they may not want to run later versions of CentOS because it just won't work for them.
- Higher probability of hardware failures over time
Well that may be likely but so is getting new hardware that fries itself.
- Performance bottlenecks
If its not had a performance bottleneck in 5 years.. its very unlikely to have one in 5 years future.
The bigger issue is stuff thats outside of an OS. A) What is the business model of the company? B) Is it expecting large growth or small growth or decline in the next 5 years? C) What are its planning mechanisms for current operations? Future operations? D) How is spending money on updates, consulting, and unknown unknowns going to lower costs versus having a known cost level (it works why play with it) E) Why and how are they using the equipment currently?
If you answer those questions and they are wanting to improve or you can show significant lowering of costs, then an upgrade may be in order. But for a lot of places those questions and how the company is run will make upgrading more expensive or more likely for project failure.
Any other thoughts?
Shawn
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Thanks to everyone for their comments so far.
The "server" in question is a basic 2 node cluster connected to an MSA500.
It runs a variety of applications including Oracle, Apache, Samba, and a proprietary app built by another vendor.
The hardware is monitored, maintained and backed up regularly.
The setup is mission critical to my client. They spent a lot of time and money to make sure it wouldn't go down.
The list's point is well taken that old *nix installs are very reliable long term. I've had similar experiences. Given this particular client's need for a reliable, stable, redundant system, I was contemplating alternatives or future upgrades rather than letting things age.
Shawn
Hi Guys,
I have a client who hopes to keep their server another 5 years making it 10 years old at that time.
At this point there are no plans to add new infrastructure or a new server to the mix. Their business model is fairly static.
I'd like to see them upgrade. Can anyone suggest specific reasons why running a business on 10 year old equipment is a bad thing?
Specific arguments I can think of would be:
- Hard/Impossible to find replacement hardware
- Lack of support for both H/W and S/W
- Possibly unable to run current versions of CentOS
- Higher probability of hardware failures over time
- Performance bottlenecks
Any other thoughts?
Shawn
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On Fri, Oct 9, 2009 at 8:01 PM, Shawn Everett shawn@tandac.com wrote:
Thanks to everyone for their comments so far.
The "server" in question is a basic 2 node cluster connected to an MSA500.
It runs a variety of applications including Oracle, Apache, Samba, and a proprietary app built by another vendor.
The hardware is monitored, maintained and backed up regularly.
The setup is mission critical to my client. They spent a lot of time and money to make sure it wouldn't go down.
The list's point is well taken that old *nix installs are very reliable long term. I've had similar experiences. Given this particular client's need for a reliable, stable, redundant system, I was contemplating alternatives or future upgrades rather than letting things age.
Shawn
Another reason might be to avoid memory/data corruption. Search Google news for a recent report from Google about how they found more memory errors than conventional wisdom has held to be expected.
However, consider this. IT has a reputation among most businesses of always wanting new toys. Many times, they have a point and it seems that IT is more interested in getting new things for no real reason. This sounds like one of those times.
The better solution would be to make sure you are prepared for when the hardware does fail. Inform the client that you understand that they don't want to upgrade the servers, and that hardware failure is not a case of "if" but "when". Lay out a plan to them describing what would happen when that occurs, and how you will make sure that their downtime is minimal.
My recommendation would be to look into VMware P2V conversion, and test it out on one of the servers if you can. Schedule a time to run it once a month or so, and make sure any data that would change is also backed up. You could back it up to a $99 1TB external USB drive, very cheap (just turn off the drive when you're done with the backup). When the servers finally die, you can bring up a new server, pop the VM onto it, and they are back up and running. You might not even need P2V if you can rsync the entire system off to the external drive. The important thing here is to make sure you test bringing up the backup system before you're in an emergency.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
thus Brian Mathis spake:
On Fri, Oct 9, 2009 at 8:01 PM, Shawn Everett shawn@tandac.com wrote:
Thanks to everyone for their comments so far.
The "server" in question is a basic 2 node cluster connected to an MSA500.
It runs a variety of applications including Oracle, Apache, Samba, and a proprietary app built by another vendor.
The hardware is monitored, maintained and backed up regularly.
The setup is mission critical to my client. They spent a lot of time and money to make sure it wouldn't go down.
The list's point is well taken that old *nix installs are very reliable long term. I've had similar experiences. Given this particular client's need for a reliable, stable, redundant system, I was contemplating alternatives or future upgrades rather than letting things age.
Shawn
Another reason might be to avoid memory/data corruption. Search Google news for a recent report from Google about how they found more memory errors than conventional wisdom has held to be expected.
For the archives, it can be found here:
http://www.cs.toronto.edu/~bianca/papers/sigmetrics09.pdf (about 300k)
http://blogs.zdnet.com/storage/?p=638
However, consider this. IT has a reputation among most businesses of always wanting new toys. Many times, they have a point and it seems that IT is more interested in getting new things for no real reason. This sounds like one of those times.
The better solution would be to make sure you are prepared for when the hardware does fail. Inform the client that you understand that they don't want to upgrade the servers, and that hardware failure is not a case of "if" but "when". Lay out a plan to them describing what would happen when that occurs, and how you will make sure that their downtime is minimal.
My recommendation would be to look into VMware P2V conversion, and test it out on one of the servers if you can. Schedule a time to run it once a month or so, and make sure any data that would change is also backed up. You could back it up to a $99 1TB external USB drive, very cheap (just turn off the drive when you're done with the backup). When the servers finally die, you can bring up a new server, pop the VM onto it, and they are back up and running. You might not even need P2V if you can rsync the entire system off to the external drive. The important thing here is to make sure you test bringing up the backup system before you're in an emergency.
Cheers,
Timo
The better solution would be to make sure you are prepared for when the hardware does fail. Inform the client that you understand that they don't want to upgrade the servers, and that hardware failure is not a case of "if" but "when". Lay out a plan to them describing what would happen when that occurs, and how you will make sure that their downtime is minimal.
For the win!
This is by far the best approach if you want to bring them along. It has to be THEIR decision, so the best way to get them to make that decision is to sit back and say "OK, if you don't want to upgrade that is fine, but we still have to make sure we are prepared for when that hardware fails, so here is what we'll do ..."
That will probably scare the crap out of them enough to change their minds :-)
On Oct 10, 2009, at 3:08 PM, Alan McKay wrote:
The better solution would be to make sure you are prepared for when the hardware does fail. Inform the client that you understand that they don't want to upgrade the servers, and that hardware failure is not a case of "if" but "when". Lay out a plan to them describing what would happen when that occurs, and how you will make sure that their downtime is minimal.
For the win!
This is by far the best approach if you want to bring them along. It has to be THEIR decision, so the best way to get them to make that decision is to sit back and say "OK, if you don't want to upgrade that is fine, but we still have to make sure we are prepared for when that hardware fails, so here is what we'll do ..."
That will probably scare the crap out of them enough to change their minds :-)
Can you get something like the average lifespan of the circuits in hours ?
There should be something saying that in the best case scenario the processor can work X thousand hours and die. Although I don't know where that information could be available :-)
-Giovanni
On Sat, 10 Oct 2009 15:14:13 -0300 Giovanni P. Tirloni wrote:
Can you get something like the average lifespan of the circuits in hours ?
I always tell people that a computer is like a light bulb -- you can't tell by looking at it how much life it has left in it. Just like a light bulb, even with identical units, some will die ten minutes after you plug them in and others will run trouble-free for years.
At Sat, 10 Oct 2009 15:14:13 -0300 CentOS mailing list centos@centos.org wrote:
On Oct 10, 2009, at 3:08 PM, Alan McKay wrote:
The better solution would be to make sure you are prepared for when the hardware does fail. Inform the client that you understand that they don't want to upgrade the servers, and that hardware failure is not a case of "if" but "when". Lay out a plan to them describing what would happen when that occurs, and how you will make sure that their downtime is minimal.
For the win!
This is by far the best approach if you want to bring them along. It has to be THEIR decision, so the best way to get them to make that decision is to sit back and say "OK, if you don't want to upgrade that is fine, but we still have to make sure we are prepared for when that hardware fails, so here is what we'll do ..."
That will probably scare the crap out of them enough to change their minds :-)
Can you get something like the average lifespan of the circuits in hours ?
There should be something saying that in the best case scenario the processor can work X thousand hours and die. Although I don't know where that information could be available :-)
The *circuits* will last more than 10 years -- things like the processors, etc. so long as they are not abused (eg cooked (overheated)). What wears out are things like power supplies (thermal stress, aging filter caps, etc.), fans (bearings), disk drives (bearings, magnatic coating). Basicically anything that normally is warm/hot or has some kind of friction situation (anything machinical / moving parts / bearings). Yes, things like processor chips do die, but as often as not, the thermal grease has dried up and/or the processor fan died and/or a case fan has died and/or the power supply went south -- any of these can take the processor out.
-Giovanni
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On Sat, Oct 10, 2009 at 2:14 PM, Giovanni P. Tirloni tirloni@gmail.com wrote:
Can you get something like the average lifespan of the circuits in hours ?
HDs and other components should have published "MTBF" - Mean Time Between Failure
Alan McKay wrote:
On Sat, Oct 10, 2009 at 2:14 PM, Giovanni P. Tirloni tirloni@gmail.com wrote:
Can you get something like the average lifespan of the circuits in hours ?
HDs and other components should have published "MTBF" - Mean Time Between Failure
But they are never realistic in terms of actual life spans. They are computed from the failures across a large number of new components in a test environment.
At Sat, 10 Oct 2009 13:44:54 -0500 CentOS mailing list centos@centos.org wrote:
Alan McKay wrote:
On Sat, Oct 10, 2009 at 2:14 PM, Giovanni P. Tirloni tirloni@gmail.com wrote:
Can you get something like the average lifespan of the circuits in hours ?
HDs and other components should have published "MTBF" - Mean Time Between Failure
But they are never realistic in terms of actual life spans. They are computed from the failures across a large number of new components in a test environment.
Of course. "Mean", as used in "Mean Time Between Failure", is a *statisical* term, and is not a 'real' number.
Alan McKay wrote:
On Sat, Oct 10, 2009 at 2:14 PM, Giovanni P. Tirloni tirloni@gmail.com wrote:
Can you get something like the average lifespan of the circuits in hours ?
HDs and other components should have published "MTBF" - Mean Time Between Failure
which is just about meaningless, unless you have a population of 1000s of drives under optimal conditions, then its a statistical mean. at that point in time, you can expect approximately half the drives to fail.
On Sat, Oct 10, 2009 at 2:14 PM, Giovanni P. Tirloni tirloni@gmail.com wrote:
On Oct 10, 2009, at 3:08 PM, Alan McKay wrote:
The better solution would be to make sure you are prepared for when the hardware does fail. Inform the client that you understand that they don't want to upgrade the servers, and that hardware failure is not a case of "if" but "when". Lay out a plan to them describing what would happen when that occurs, and how you will make sure that their downtime is minimal.
For the win!
This is by far the best approach if you want to bring them along. It has to be THEIR decision, so the best way to get them to make that decision is to sit back and say "OK, if you don't want to upgrade that is fine, but we still have to make sure we are prepared for when that hardware fails, so here is what we'll do ..."
That will probably scare the crap out of them enough to change their minds :-)
Can you get something like the average lifespan of the circuits in hours ?
There should be something saying that in the best case scenario the processor can work X thousand hours and die. Although I don't know where that information could be available :-)
-Giovanni
You are thinking too technically-minded. This is a business decision, not a technical one. Numbers on how long components last are irrelevant to this conversation. The only point that needs to be made is that a failure can be expected, and to plan business operations around that.
Another point in this is that no matter what they do, there is a cost to maintaining an infrastructure. Buying new servers does not eliminate the risk of hardware failure, and doesn't necessarily delay it either. These kind of plans should be in place with the new or old servers.
Also, when I say "describe what would happen", I'm not talking about how a board might blow a capacitor, I'm talking about how you will react and what the impact is on the business during that time.
Brian Mathis wrote:
Also, when I say "describe what would happen", I'm not talking about how a board might blow a capacitor, I'm talking about how you will react and what the impact is on the business during that time
double or triple your support contract rates for systems deemed past EOSL, AND have no guarantees, just best effort.
another side effect of hanging onto old legacy business systems, it can become increasingly hard to migrate the data from the old software system to a new one the older they get. and if they wait til the legacy hardware has completely failed, it can become even harder. throw in things like old backup media formats (i'd hate to have to find a reader taht could read QIC tapes from a SCO system written by some funky proprietary business software written 25 years ago). Its MUCH easier to migrate to a new system if the old system can be run in parallel during the transition period.
Brian Mathis wrote:
On Sat, Oct 10, 2009 at 2:14 PM, Giovanni P. Tirloni tirloni@gmail.com wrote:
On Oct 10, 2009, at 3:08 PM, Alan McKay wrote:
The better solution would be to make sure you are prepared for when the hardware does fail. Inform the client that you understand that
<snip>
This is by far the best approach if you want to bring them along. It has to be THEIR decision, so the best way to get them to make that decision is to sit back and say "OK, if you don't want to upgrade that is fine, but we still have to make sure we are prepared for when that hardware fails, so here is what we'll do ..."
That will probably scare the crap out of them enough to change their minds :-)
<snip> Got it: what you need to do is set up a meeting to prepare a disaster recovery plan. Most business have, or are willing to look at them (at least post-Katrina). Make sure that they know just how *long* it will take to bring new systems online if there's a fire, or flood, or hurricane, or twister....
Then bring up the age of the servers and maintenance. As part of the plan, of course, spec out replacement hardware, and any performance increases it would bring....
mark
mark
On Sat, 2009-10-10 at 14:08 -0400, Alan McKay wrote:
The better solution would be to make sure you are prepared for when the hardware does fail. Inform the client that you understand that they don't want to upgrade the servers, and that hardware failure is not a case of "if" but "when". Lay out a plan to them describing what would happen when that occurs, and how you will make sure that their downtime is minimal.
For the win!
This is by far the best approach if you want to bring them along. It has to be THEIR decision, so the best way to get them to make that decision is to sit back and say "OK, if you don't want to upgrade that is fine, but we still have to make sure we are prepared for when that hardware fails, so here is what we'll do ..."
That will probably scare the crap out of them enough to change their minds :-)
This is absolutely correct. I would try and ease them into understanding that hardware does fail and that they really do need to be prepared. Once they grasp that concept they will most likely let you make the backup and after the first major fail of their existing platform, never migrate back.
Morning,
On 10 Oct 2009, at 17:12, Brian Mathis wrote:
The better solution would be to make sure you are prepared for when the hardware does fail. Inform the client that you understand that they don't want to upgrade the servers, and that hardware failure is not a case of "if" but "when". Lay out a plan to them describing what would happen when that occurs, and how you will make sure that their downtime is minimal.
Can't agree more. You want to present to the client that you care first and foremost about their business needs. It's a simple calculation - work out what's involved in maintaining the old systems, and compare that to the cost of upgrading. These sorts of conversations are often about the difference between operational expenditure and capital expenditure - your client may be willing to pay more long term support costs, and not be prepared to buy new hardware and pay you to upgrade the systems.
It sounds to me that you want to upgrade because you're assuming it's the right thing to do. That's a big assumption - go through the business cases, and make sure your client understands you're on their side which ever way the decision goes.
S. -- Stephen Nelson-Smith, Technical Director, Atalanta Systems Ltd, http://www.atalanta-systems.com
Assuming that you are going to lose this battle, If you have a virtual environment, maybe you can do a P2V for backup purposes, so if it fails, you have a backup.
If it makes you feel any better, we still have some Windows NT systems still running on original hardware. We are all terrified to touch it and we can't back it up, but they refuse to upgrade it or get rid of it.
-----Original Message----- From: centos-bounces@centos.org [mailto:centos-bounces@centos.org] On Behalf Of Stephen Nelson-Smith Sent: Monday, October 12, 2009 2:19 AM To: CentOS mailing list Subject: Re: [CentOS] 10 Year old IT Infrastructure
Morning,
On 10 Oct 2009, at 17:12, Brian Mathis wrote:
The better solution would be to make sure you are prepared for when the hardware does fail. Inform the client that you understand that they don't want to upgrade the servers, and that hardware failure is not a case of "if" but "when". Lay out a plan to them describing what would happen when that occurs, and how you will make sure that their downtime is minimal.
Can't agree more. You want to present to the client that you care first and foremost about their business needs. It's a simple calculation - work out what's involved in maintaining the old systems, and compare that to the cost of upgrading. These sorts of conversations are often about the difference between operational expenditure and capital expenditure - your client may be willing to pay more long term support costs, and not be prepared to buy new hardware and pay you to upgrade the systems.
It sounds to me that you want to upgrade because you're assuming it's the right thing to do. That's a big assumption - go through the business cases, and make sure your client understands you're on their side which ever way the decision goes.
S. -- Stephen Nelson-Smith, Technical Director, Atalanta Systems Ltd, http://www.atalanta-systems.com
_______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Shawn Everett wrote:
Thanks to everyone for their comments so far.
The "server" in question is a basic 2 node cluster connected to an MSA500.
It runs a variety of applications including Oracle, Apache, Samba, and a proprietary app built by another vendor.
The hardware is monitored, maintained and backed up regularly.
The setup is mission critical to my client. They spent a lot of time and money to make sure it wouldn't go down.
Will you still be able to get hardware support and repair parts 10 years out? And even if you can, you may reach the point before 10 years where it would be cheaper to replace than maintain. If you are paying for support, that's just going to cover repair/replacement of the nearly obsolete component - and newer equipment takes a lot less power for the same capacity.
The list's point is well taken that old *nix installs are very reliable long term. I've had similar experiences. Given this particular client's need for a reliable, stable, redundant system, I was contemplating alternatives or future upgrades rather than letting things age.
I'd consider 5+ years to be a reasonable expectation - 10 is possible but a stretch. You'll definitely be exposed to new security vulnerabilities that may be discovered beyond the update support cycle on the OS so it doesn't fit in any sort of "best practices" concepts, especially if there is any internet/remote exposure.
Specific arguments I can think of would be:
- Hard/Impossible to find replacement hardware
- Lack of support for both H/W and S/W
- Possibly unable to run current versions of CentOS
- Higher probability of hardware failures over time
- Performance bottlenecks
Any other thoughts?
Shawn
__
Shawn,
i dont think you mentioned the specific hardware involved...
i.e. brand and model number and config
it makes a difference in how we would approach it...
mainly because we have some 10 year old and older hardware that has been running rock solid it's entire life and we expect several more years out of some of it...
we keep hot and cold spares of everything though...
to be semi generic, i am talking about business / industrial rackmount Compaq & HP servers, and some telco quality Cisco of course...
:-)
please do share...
- rh
Approach him like this. Tell him if he plans on moving his business forward within the next 5 years he should think accordingly.
On Sun, Oct 11, 2009 at 11:26 PM, R-Elists lists07@abbacomm.net wrote:
Specific arguments I can think of would be:
- Hard/Impossible to find replacement hardware
- Lack of support for both H/W and S/W
- Possibly unable to run current versions of CentOS
- Higher probability of hardware failures over time
- Performance bottlenecks
Any other thoughts?
Shawn
__
Shawn,
i dont think you mentioned the specific hardware involved...
i.e. brand and model number and config
it makes a difference in how we would approach it...
mainly because we have some 10 year old and older hardware that has been running rock solid it's entire life and we expect several more years out of some of it...
we keep hot and cold spares of everything though...
to be semi generic, i am talking about business / industrial rackmount Compaq & HP servers, and some telco quality Cisco of course...
:-)
please do share...
- rh
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
R-Elists schrieb:
Specific arguments I can think of would be:
- Hard/Impossible to find replacement hardware
- Lack of support for both H/W and S/W
- Possibly unable to run current versions of CentOS
- Higher probability of hardware failures over time
- Performance bottlenecks
Any other thoughts?
Shawn
__
Shawn,
i dont think you mentioned the specific hardware involved...
i.e. brand and model number and config
it makes a difference in how we would approach it...
mainly because we have some 10 year old and older hardware that has been running rock solid it's entire life and we expect several more years out of some of it...
we keep hot and cold spares of everything though...
to be semi generic, i am talking about business / industrial rackmount Compaq & HP servers, and some telco quality Cisco of course...
:-)
please do share...
- rh
Yeah, if it's an IBM mainframe, I'd give it another 10 years, easily. Spareparts will probably still be available long after my death. But don't look at the maintenance-bill you get from IBM. At least not before breakfast. It's not for the faint-hearted or those with a weak stomach....
;-)
Rainer
-
Alan McKay -
aurfalien＠gmail.com -
Benjamin Donnachie -
Bill Campbell -
Blackburn, Marvin -
Brian Mathis -
Drew -
Frank Cox -
Giovanni P. Tirloni -
Gordon McLellan -
James Matthews -
John R Pierce -
Les Mikesell -
mark -
R-Elists -
Rainer Duffner -
Robert Heller -
Shawn Everett -
Stephen John Smoogen -
Stephen Nelson-Smith -
Tait Clarridge -
Timo Schoeler