Well I agree that it's ridiculous design to have GbE NICs on PCI in this age.
I noticed that comment the other day and made it a point to never buy one of those
based on that detail.
I also seem to have noticed (check this) that the IP35-E lower end unit may not have
the same deficiency in that respect, and it seemed (from the thread dedicated to it,
the IP35-E review thread) to be a decent low-cost P35 motherboard.
That being said I went with what seemed to be the best-of-breed at the time,
the P5K-Deluxe when I built my first fairly high end system about 4 months ago.
It's been FINE except for a few usual BIOS bugs that you can work around but
which are annoying, but I think EVERY motherboard has a few of those if you're really
a "power user" and expect to demand a lot of unusual configurations to work.
For my second mid-high end build I went with the lower cost P5K-E, there are only
minor differences between the P5K deluxe and P5K-E and I didn't feel any of them
would be relevant to me enough to spend $80 more on a 2nd deluxe model.
Mostly it lacks some of the extra cooling, but I run in a very good case with great
airflow and use high end heatsinks for the CPU and knew I couldn't and wouldn't want
to overclock to the point where the chipset would get too hot anyway.
Like you I want nearly perfect stability, reliability, and a modest overclock that's
basically "low hanging fruit", a "free" upgrade with no impact on stability.
So basically it's one of those two models that I'd suggest to you, I consider them
superior to the IP35-Pro based on their better GB-LAN design and some of the ASUS
features of the BIOS and their motherboard cooling / heatpipe capacities, the
multi-phase CPU power system, etc.
Sure there are even BETTER higher end motherboards in the X38 chipset series,
or even the P5K Premium, but I don't see any added stability or performance
advantages that are crucial in anything more expensive.
My main gripe about the P5K's below the Premium are the Mosfets on the rear
of the motherboard that get kind of hot, I'd have put them on top and cooled them
better. But it evidentally isn't a failure mode of the motherboard that these run
hot, and people into extreme overclocking with even higher end CPUs than I have
get them a LOT hotter / more heavily taxed than I do, so I feel safe though it
is a compromise.
If you want to look at the whole architecture of the PC in general it's full of half-baked
compromises like the whole case design -- horrible airflow / heat management,
difficult to assemble, difficult to service, collects dust, needs tools to work on it vs.
tool-less modular workstation / server type designs, etc.
Yes $50,000 workstations from SUN or IBM whatever are going to have great
SUPPORT / field service, they may have better mechanical designs, better
electrical engineering, better quality control, etc. then again many workstations ARE
now based on Intel processor type of PC architecture, and many DO take PC
add in cards etc. so some of those design advantages are going away not because
PCs are getting "better" but workstations are getting WORSE using commodity parts.
But if you choose decent quality components wisely, assemble them CAREFULLY
(i.e. actually USE the grounding wrist-strap everyone laughs at, etc.), and maintain
them reasonably (DUST it out monthly if it's not in a very dust-free environment,
control the humidity, keep pet hair out of it, don't smoke and get the whole thing
grimy, etc.), PCs can and will last 5-10 years without any more than perhaps
replacing one or two parts that may fail due to old age. Usually you can use them
until they're so "slow" or so incompatible with new desirable technologies that
you sell them / donate them in full working order just to upgrade.
I don't think the MTBF (reliability lifetime expectation) of mid-range workstations
are mugh better than the 5+ year relatively problem free lifespan of a carefully built PC.
Sure you may have a field service contract option for the workstation that will get
any and all of it swapped out within 1 business day on-site service, but that's about
the main advantage. Things WILL fail statistically ABOUT as often with either, the
question is what will be the timeliness of diagnosing/fixing the hardware or software
fault.
For the USER though the workstation type SUPPORT doesn't do much good in the
MOST cricital area -- YOUR OWN DATA because even if it's a cheap and quick
2 hour down-time deal to replace a dead hard disc, nobody but you can actually
re-load all your applications and data and customizations to get the repaired system
back to useful service with no data loss once you have a working machine + OS ready
for a restoration. That's the big down-time risk you MOST carefully have to mitigate.
Given that, I'd say that two or more offline hard discs like external USB / firewire
discs are pretty mandatory backup devices; store at least ONE offsite in a secure
fire-proof place like a safe-deposit box. Do full backups weekly, partial ones daily,
keep the last N months of backups in a rotation just in case. Obviously have a plan
to archive the last N years of financial data for tax/record purposes etc.
Have a system to "ghost" the main operating system and application data so that
if your drive dies you can be 100% back online without needing to reinstall the
OS or applications from a drive-imaged backup.
Use a 4-drive RAID for your main operating system and application disks so that
any TWO of those drives can fail or corrupt and you have ZERO downtime due to that,
just plug in replacement units when you're ready to shut down for the day and
you're back in business with redundancy. Keep 2 spare drives of the same models
FOR such replacement not IF but WHEN needed.
Other workstation advantages? Well software licensing/locking and availability often
forces people to use only certain SUN, HP, IBM, whatever models just because you
CAN'T get/run the apps. you need for PCs or MACs. Fortunately this is less commonly
the case now. The "node locking" of certain high end applications MAY be a problem
for you in terms of downtime though -- if your MAIN PC's motherboard or hard drive
dies, and your applications are serial number LOCKED to run on only those hardwares,
you BETTER know how to get that updated QUICKLY or you will be DOWN until you do.
If you can't tolerate even a few hours of downtime, there's another issue which is
the PC (or workstation!) itself. If it DOES blow up, flood, catch fire, GET STOLEN,
what then? The only real solution here is to have a second INDEPENDENT PC
all loaded up with your software READY TO GO needing only to synchronize with
your latest daily backup of your files to get it 100% up to speed for switch-over use.
Obviously you'd have to keep that unit in a different secure location for full security,
though just having a second PC in the office/home will at least give protection
against downtime due to a fault with the first one.
So yeah, I'd go with the PC, any mid-high-end enthusiast motherboard like P5K-E or
above should be FINE, no benefit of a server motherboard or "workstation" unless your
software NEEDs the extra RAM or multiple-CPU or vendor hardware / OS
available only in those.
Spend the money you save on backup stuff like external drives, network attached
storage, backup software, off-site backup / secure storage / off-site hosting / whatever
for data security issues.
Gb Ethernet -- actually it IS a big deal (TO ME) to be able to SATURATE the
LAN because when I do things like backups from PC to PC LOCALLY or use
certain distributed computing applications, I can and DO get 100MB/s traffic
rates over the GbEthernet. A good RAID will give you 50-100MB/s READ
performance, so that WILL use up GbE full bandwidth. So I agree it's wise
to NOT get the crippled solution IF you will have any similar PC-to-PC LAN
based file-server/backup-server/SAN type needs.
Facebook
Twitter