Oh, Jeez! My condolences! I won't let SWMBO even *sit* at any of my workstations! Need something? OK, I'll build you a laptop with that for you to use (she currently has two such laptops in addition to her workstation).
[Her replacement workstation is sitting in the living room waiting to be built -- as is her replacement laptop. Her old workstation will get discarded (pull and sanitize disk) and one of "my" laptops will be restored to its original image -- the other discarded. Just need ten or twenty more hours in the day... EACH day! :< ]
The laptops I build for students restore in less than 5 minutes. But, there's just Windows, some tools and MSOffice, there. The point being it is really NOT very painful to do a restore -- so, it becomes a viable way of working through any problems!
I had one lose the boot sector to a hardware failure. Simply would not restore. And, apparently the sector remapping software wouldn't accommodate it.
No problem. Slap another disk in and restore the image to that!
My "live repository" sits, in duplicate, on multi-TB spindles. (I have a *lot* of stuff archived!)
I have offline copies on optical media (which have to be treated gingerly lest they give up the ghost).
And, really precious things are on MO and tape. It may take me a while (a LONG while) to recover a lost spindle. But, I *can* recover them!
I have several COTS NAS's that I spin up, as needed, so I can "push" a copy of a work-in-progress off to another spindle if I fear I may screw something up with the local "working copy".
Biggest hassle is retrieving big objects from the archive. E.g., when I build multimedia presentations, I often have to pull hundreds of GB of audio/video "templates" off the archive to explore, "live". Even at gigabit speeds, it takes a fair amount of time!
Biggest *headache* is keeping track of "little projects" that may have been pulled out for a bit of touchup -- and then forgetting which copy has the most current set of changes.
E.g., I spent a few hours revising a paper pulled onto a laptop while sitting in car waiting for SWMBO. If I forget to merge those changes back into the original, I may find myself wondering "Didn't I change all this, already?" at some future time...
I fixed a lawn mower for some folks, one time. The stop switch wasn't working, so I put on auxiliary stop button, that goes on the mower push handle. Couple weeks later, the user (teen age boy) can't get it to start. I told him it would start a lot easier if he'd take his finger off the kill button while he was pulling the start cord.
Yeah - that's what I have come around to: once have the procedure down pat, a re-image is just too simple and takes too few clock hours (actually minutes) for me to even consider anything else.
The whole business with Windows' backtracking the registry to "Restore Points" becomes, IMHO, sort of silly - especially because then I have to trust that process and assume that nothing else is amiss.
And, for the insufficiently-paranoid who may only have one backup device: consider the scenario where a flaky USB card or some other weird failure fries drives as they are connected.
Once I cooked 2 or 3 backup drives before it dawned on me what was happening. Since then I make it a point to always keep at least one drive where I cannot get to it too easily - in hopes that I will come to my senses soon enough in a similar situation.
The updating takes a long time, so I just start it, last thing in the day and let it run overnight.
Right now, perhaps thanks to Oren's prompting, I am finally creating a Win7 slipstreamed DVD. The one I have now has SP1 but none of the subsequent updates. The one I'm creating now will have about 200 updates included.
If I am moving tools to a new machine (not adding any "new" software or "updating" anything) *and* I have all of the drivers for that machine already downloaded, it takes me about 3 full *days* to build one of my workstations. (I have three; and the applications on each are essentially different).
But, at the end of that process (ordeal), the machine is "set for life" (I don't add stuff to a workstation once it has been built and configured).
At the end of each calendar year, I reevaluate my tools/equipment and upgrade (out with the old, in with the new) as I deem appropriate. I am *really* reluctant to replace/upgrade a workstation in that process (typically, the machines are faster than my meatware so why incur the cost and inconvenience of an upgrade unless there is some HUGE advantage to doing so?). And, only mildly less so willing to upgrade a piece of software ("for a NEW set of bugs!").
I chuckle at folks who are always buying the latest and greatest. Unless they're playing video games, they're just throwing money (and time!) away...
Exactly. Entropy is very apparent in any MS system!
For students, having to "live without" the machine while it is being "repaired" is a big disincentive to NEEDING it repaired.
IME, people would much rather the convenience of being able to have a "working" machine than they would fret over the potential of "losing something" (and needing a manual "recovery and rebuild"). Esp if you can give them a way to keep their "precious stuff" someplace that isn't corrupted/lost in the process.
[Of course, this also works great for *me* -- as I have no desire to be recovering lost files for all sorts of careless users!]
Question: has performance of PCs available on an common peasant's budget progressed to the point where I would notice a diff if I upgraded from my current GigaByte Z87X-UDH4 / i7-4770K @3.5 GHz with
I had that happen on one of my servers. Kept wondering why I was having problems with USB devices (e.g., external disks). Swapped out USB hubs, cables, etc.
Of course, I'd been deligently ignoring the "warning light" as it was a hassle to hook up an IPMI client to query the machine ("What's wrong?").
Discovered a bad USB/FW card was the cause of all the problems. Pulled that and everything played nice together.
I keep backups on different spindles on different machines. My newest approach connects external USB drives to little headless machines (Optiplex FX160's) and mounts them remotely. Pull whatever files I want over the network -- knowing that there is nothing else "running" on the FX160's that could hiccup.
Other copies of the files on any particular volume reside on other disks. I have a database that tracks where each is located (disk, directory, filename -- a file can have a different name and still be the same file!) along with information that helps me verify its integrity (checksum, date, etc.).
Another program just walks through each disk's contents (on the FX160's) verifying the checksums of each file against the STORED checksums in the (remote) database.
So, if something is remiss, I get notified that the file is experiencing bitrot and I have to arrange for a "backup copy" to be mounted so the rotting copy can be restored.
[I don't want this to be an automatic process as a hiccup could wipe out ALL copies of a file!]
This ensures that copies exist on independent media, controlled by independent CPU's, is *verified* regularly (just HAVING a copy doesn't mean that copy is still accessible!) and can be LOCATED easily (with literally MILLIONS of files in my repository, there's no easy way to KNOW where a copy of a particular file might reside: which USB disk? which CD? DVD? etc.)
Not really. I just spend a lot of time in front of machines actually "doing real work" (not watching movies or browsing the web, etc. but, rather, authoring multimedia presentations, designing circuit boards, writing software, etc.). In each of these cases, *I* am the limiting factor. A faster machine just means the machine spends more time WAITING for me to figure out which key I want to press next!
I would seriously doubt it. Unless you're doing gaming, etc. And, in that case, the video card is more important than the CPU (video cards have GPU's on them that do most of the graphic "processing")
I use a 1.8GHz Core2 Duo for our HTPC -- and that probably has the most demanding "real time" work to do (you don't want a movie to start skipping, pixelating, etc. while watching).
There are many things that my workstations may spend *hours* crunching (rendering 3-dimensional models, generating videos of those models moving in a predefined "scene", compiling thousands of files for a project I'm working on, etc.). But, having a machine that is *10* times faster would cut that down to *an* hour (or two) -- still too long for me to be sitting there twiddling my thumbs!
So, I start a machine on one of those lengthy tasks and then find something else to do while it is "busy thinking" (either using another machine *or* on that machine WHILE it is thinking).
My current big bottleneck is speed of network fabric (1Gb) and slow USB2 interfaces (I've not bothered to move to USB3 as it means replacing a lot of hardware JUST to get faster disk interfaces).
When I started out in this business, I could make exactly *two* changes to a project in an 8 hour "shift" -- 4 hours to see what the results of my change would be! The machines were *so* slow...
As a result, I learned how to plan the use of my time around the capabilities of the machines. As the tasks that I now do take considerably more machine power than back then, it's just not possible to buy hardware that is fast enough to ensure "no waiting" (at least not for the things that *I* do).
So, I rely on the same sorts of skills I learned decades ago and PLAN how my time will be spent so I'm not idle, waiting for a machine.
When I first started doing 3D CAD drawings on 25MHz 386's -- when THAT was the "spare no expense" hardware available -- it would take a machine 24-26 hours to render *one* drawing. If the lights blinked, your heart sank! (Oh, crap! I wonder if the machine "saw" that??)
Now, I can render that same drawing in a minute or two. But, now want that drawing to "move"... I want to see the mechanism operate, see the shadows that are cast as it moves around relative to the artificial light source, etc.
So, I'm back at the day-long renders :>
But, unlike back then, I don't have to move to another machine in order to keep working!
One of my granddaughters got a BS in computer graphics from Drexel and now she makes a decent living doing same.
One of the companies she worked for as a student intern said she "revolutionized the way they do business" for them - by installing services on all of the employees' PCs that allowed offloading of rendering tasks to those PCs - creating a multi-PC rendering farm so productions could be rendered in minutes/hours instead of days.
Dunno how many PCs it was... but I would guess plenty...
Called a "network of workstations"/loosely-coupled cluster.
The amount of "surplus" computing power "being wasted" is staggering. Ages ago, I recall walking by the stock room with an employer and he glanced up, sighed and said, "Think of all that power sitting there *wasted* (bare chips)..."
When I designed the home automation system, I was very aware of this as a means of cutting cost 9at the expense of complexity!) by leveraging "underutilized" computers that were *needed* in particular places (in order to interface to particular devices) -- but that spent
99.9999% of their time twiddling their thumbs: "Why not use that surplus capacity INSTEAD OF adding a big computer someplace that throws off lots of heat, has noisey fans, etc.?" So, the design is far more complex. And, in many ways far less *efficient*. But, if stuff was being "underutilized", then who cares about efficiency?!
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.