Long Term System Stability?

[jaclaz]

@NoelC

As said, performance differences may be trifling on today's fastish devices, but try recovering a corrupted database (or filesystem) in which entries/contents are not contiguous (or compacted, or defragged), then come back saying that it is unneeded (again with moderation).

 

In a "perfect world" where you have TWO (daily) backups the above doesn't obviously make any sense, as you won't EVER be needing to recover anything from anywhere.

 

If you can convince everyone to adopt such a backup strategy, then all is fine and dandy.

 

BTW, by the same token, aligning filesystems (or databases) to device blocks makes little sense on fastish machines/devices, but still on slowish ones it is *needed*.

 

@TELVM

For whatever it matters, the same can be said about conventional hard disk manuifactured in - say - last 15 years (the fact that where the data actually is has nothing to do to where Windows "sees" it being), think a bit about the G-Lists, P-Lists and spare sectors.

The advent of SSD's has only exacerbated the same "problem" or "feature" (Windows or any OS, with the exclusion of the proprietary on-board firmware of the disk have NO idea of where actually the data is).

As a matter of fact *any* chip based device (such as, besides SSD's the common SDcards and USB sticks) stores data not only "wherevet it sees fit", but also in a particular form/pattern (including where appliable CRC and/or other forms of error correction)  that makes "chip off" reading particularly "nasty".

 

jaclaz 

[MrMaguire]

With everything that I've been seeing, experiencing and reading concerning SSD drives, I have decided that I prefer hard drives. Call me old fashioned, but I also prefer tapes and floppy disks over optical media and flash memory based media.

 

@TELVM

 

Yeah, it's a BTX system from 2005, with a Pentium D and I haven't managed to get some temperatures from SpeedFan or Core Temp yet. I've seen someone do the same modification to an Optiplex GX620 tower, except they didn't cut out the grill. To me, cutting out the grill for a little bit better airflow seems unnecessary and destructive.

[NoelC]

...TWO (daily) backups...

If you can convince everyone to adopt such a backup strategy, then all is fine and dandy.

 

 

Funny thing, but right now, today, with Windows Backup (wbadmin) and a basic USB2 external drive, my nightly backups complete in an hour or two.  Most importantly, when I do end up working late, I really can't tell that a backup is going, save for a flashing light on the external drive and maybe some sub-audible seeking sounds from it.

 

It's truly "set it and forget it" technology already.

 

I looked earlier today:  I have 54 system images on the drive available to choose from should I need to restore something.  About 2 months worth of daily backups.

 

If I did the same backup twice daily, say at 12 hour intervals, would there be the problem?  No.  Since fewer blocks of data will be changed in 12 hours the backups would complete more quickly.  And given that VSS-integrated System Image backups only store differences, I imagine the number of available backups from which a restoral could be done would still be in the 1 to 2 month range.

 

I'm not likely to change my system image backup to twice daily, personally, since I have other data backups that run at various times (I back critical data up 4 different ways automatically).

 

I suspect that a great deal of the reason that people don't do backups is 1. psychological and 1a. resistance to getting geeky enough to understand it thoroughly and take the time to set up a good backup strategy, which not only involves setting things up, but also testing to ensure one understands what would be needed when restoring data, and to see that the backed-up data really is sound.  Again, VMs make a great test environment for this.

 

-Noel

[NoelC]

Yeah, it's a BTX system from 2005, with a Pentium D and I haven't managed to get some temperatures from SpeedFan or Core Temp yet.

 

Not sure what your goals are, but for pure monitoring try out HWMonitor...

 

 

-Noel

[jaclaz]

I suspect that a great deal of the reason that people don't do backups is 1. psychological and 1a. resistance to getting geeky enough to understand it thoroughly and take the time to set up a good backup strategy, which not only involves setting things up, but also testing to ensure one understands what would be needed when restoring data, and to see that the backed-up data really is sound.  Again, VMs make a great test environment for this.

Sure and to this you may add that the relevance of it is never taught to anyone (of the kids) while most of the experienced folks (but not geekish enough to "evolve") have been scolded (besides by some data loss because they didn't have a backup ) also by the (generically) lack of reliability of common backup tools and/or backup media or by the extreme complexity of the actual steps involved.

 

Nowadays backing up (twice) on hard disks and doing periodical upgrades/replacement of the disk devices is easier, faster and less complicated, yet everyone is commonly convinced that it is "optional" and not really-really needed.

 

jaclaz

[MrMaguire]

@NoelC

 

Thanks for the software recommendation. I tried it and it doesn't show a CPU temperature. Just like SpeedFan and Speccy, it shows hard drive and GPU temperatures only. Core Temp just shows 0c.

 

By the way: Happy post count! (666)

[TELVM]

Long story short, I'd never defrag an SSD. In fact I want it heavily fragmented, to capitalise on multichannel parallelism and optimise wear levelling.

 

... To me, cutting out the grill for a little bit better airflow seems unnecessary and destructive.

 

Merciless destruction is exactly what grilles deserve    . Look at the damage they inflict on airflow:

How do different fan mesh patterns affect fan and chassis airflow?

Effects of Grill Patterns on Fan Performance/Noise

Me, I make no prisoners amongst grilles:

[MrMaguire]

I'm thinking something more like this:

 

 

 

Interesting chart though.

[jaclaz]

Long story short, I'd never defrag an SSD. In fact I want it heavily fragmented, to capitalise on multichannel parallelism and optimise wear levelling.

 

And, even shorter, NOONE actually suggested you to defrag one.

At least myself and MikeRL were talking of defragging compacting the Registry.

 

jaclaz

[TELVM]

I'm thinking something more like this:

 

 

 

^ Yep that'd improve case ventilation a lot.

 

If there must be a grill, it's better to place the fan upstream of it. Grilles are less nocive when they are in the relatively higher pressure zone downstream of the impeller blades.

 

But the only good grill is a dead one, no mercy    .

 

[NoelC]

Out of curiosity, what is the grill supposed to keep from traversing the case barrier?

 

• Keep bugs out of the system?
   
• Keep explosion fragments contained?

 

My systems and fans run 24/7 anyway, and my office isn't terribly infested so I doubt many bugs are going to make it in the back.

 

Plus I get everything out again occasionally.  Ever blow accumulated dust out of your system with a leaf blower? 

 

-Noel

[koganstyle]

The most reliable machine I ever came across was a novell netware 3.1 server that had not been rebooted in 8 years LOL. As for windows I’ve never seen a machine that lasts long term, maybe industrial pc's that I’ve setup with a scheduled reboot that no-one uses (Windows 7 by far the best). Windows 8 seems to be slightly better but I don't have enough people running it yet to comment Intelligently. I think a major factor in stability seems to be the structure of the "Windows Profile" as you can see MS has been tinkering around with it for years. I spend a lot of time troubleshoot corrupt windows profiles.

[TELVM]

... what is the grill supposed to keep from traversing the case barrier?

 

Our fingers. Case makers don't like law suits such as "your case fan chopped my pinkies!", and most people don't know how nocive grilles are for airflow and hence don't give a ****.

 

The best cases for airflow have raised concentric rings grilles (the less damaging type), which are also easily detachable:

 

[NoelC]

Summarizing the discussion on fans and grilles, I think it's safe to say that a system with excellent cooling capacity is likely to be more stable than one that experiences wide ranges of temperature fluctuations.  Heating and cooling cycles are hard on electronic hardware.

 

Good point, TELVM, thanks for bringing it up.

 

This is another good reason to use SSDs...  They don't make nearly as much heat as HDDs.

 

One of the things I'm fond of in Dell's Precision T5500 workstation is that it was built with a bunch of excess cooling capacity.  I can start a 3D render, and it's actually a good part of a minute before the fans even pick up speed off their idle levels, since the heat from momentary power increases is easily absorbed into the cooling towers, etc.  It's a nice, quiet system for normal use.

 

-Noel

[TELVM]

... Heating and cooling cycles are hard on electronic hardware ...

 

^ This. The narrower the Δtemp between idle and full burner, the lesser the punishment on the electronics.

 

Also most components' life expectancy is shortened dramatically by higher temps, lytic caps are a good example:

 

105C rated:

@ 105C - 1000hr

@ 95C - 2000hr

@ 85C - 4000hr

@ 75C - 8000hr

@ 65C - 16,000hr

@ 55C - 32,000hr

85C rated:

@ 85C - 1000hr

@ 75C - 2000hr

@ 65C - 4000hr

@ 55C - 8000hr

 

 

One of the best articles I've ever read about cooling comps by air:

The Big Air Cooling Investigation