Mr. Jetson teaches network technology



  • @atazhaia said in Mr. Jetson teaches network technology:

    There was also no rule demanding that users must change passwords regularly (error nr 2).

    From the NIST:

    0_1518659288016_Screenshot 2018-02-14 at 19.47.52.png


  • Java Dev

    Mr. Jetson also has a book about Network Administration with Windows Server 2012 (2013 edition). While Network Technology was basic server administration, this is advanced server administration. And instead of continuing with Windows Server 2008, we must use the latest and shiniest for this book. Note also how Mr. Jetson had time to learn Windows Server 2012 in less than a year for this book, but could not be arsed to update his Linux chapter to Ubuntu 16.04 for the 2017 version of that book. Priorities!

    The first part is entirely dedicated to all the amazing new features of Windows Server 2012 as compared to 2008.

    You activate Windows under Windows Activation in System.
    The program System Information may be useful.
    System Configuration is another useful program.

    Yes, this is all he had to say about these programs. First channeling Captain Obvious for the first point and then following with saying nothing about the other two. What are they used for? Fuck that! They're useful, that's all you need to know.

    BSD licenses, for example FreeBSD, is a type of Open Source. With this license the user can modify the program and sell it. Without sharing the source code. You can therefore use an Open Source program and use that in a program or operating system that you then sell for money. This is done for example in MAC OSX which is built upon a version of FreeBSD Linux.

    :facepalm:

    Free content. An author can revoke the rights of a product and make it public, Public Domain. Creative Commons, which is used by for example Wikipedia is an example of that.

    Have you bothered looking up how the Creative Commons license actually works? Because CC != public domain.

    Then there's a very brief chapter on PowerShell and scripting. Does it say how to use PowerShell or make scripts? Well, as Mr. Jetson cba to learn how to do that you can't either. It's part of the course to learn scripting, just that Mr. Jetson skips that and just says that "in some cases after running a guide you can export the script and run that instead for future installs". Does he say how to actually save the script in the proper format or how to run it? Nope! All you get is a list of common aliases without explanation and that you can use the cmdlet get-command to list the available cmdlets. Helpful.

    A harddrive in MS-DOS and Windows 3.1-98 can contain a maximum of two primary partitions or one primary and one extended. In Windows NT 4.0 and 2000/2003/2008/2012 you can have up to four primary partitions or three primary and one extended.

    Hmm... Iirc all those OS handle the latter system just fine, as it's in the MBR standard.

    [Regarding a broken drive in a RAID1.] As soon as possible it should be replaced with a new. To do that you need to shut down the computer, however.

    If you need to shut down the server to replace a drive in a RAID that's a very shitty server. (Not counting a repurposed home PC here.)

    RAID is for Redundant Array and Independent Disks

    You can also create a RAID system in software ... but this is no real RAID so we'll skip that in this book.

    Problem with the short-term memory there? You spent the last chapter detailing how to set up a software RAID in Windows! He also keeps repeating himself so at the start of the following chapter he explains yet again what RAID is in case we forgot it.

    Today in servers disks are connected with SAS. The difference between those and the cheaper S-ATA drives is that SAS is a lot faster. ... The SAS connector has a maximum transfer speed of 6 Gbit/s, the same as S-ATA III.

    Yep. A lot faster, as you said.

    Older drives could run at 5400rpm or 7200rpm at best. Today a S-ATA drive commonly runs at 7200rpm or 10000rpm. A SAS-disk commonly has 7200, 10000 or 15700rpm.

    Ah, yes, the common consumer drive speed of 10000rpm.



  • I think we've established quite thoroughly that "Ignorant author is ignorant".


  • Java Dev

    @steve_the_cynic To his defense, he and I do agree on one point: that virtualisation is an important topic that needs to be brought up and properly taught, even if it's not in the course goals. And as long as it's Windows Server with a GUI he can give correct information and instructions for how to use it.

    Otoh, the errors range from harmless to serious, and poking fun at them is fun. Although I am pretty much out of books now. There is one more that I have: Digital communications, but it has a lot of copy-pasted chapters from the other books so there's actually not that much new stuff in it.


  • Discourse touched me in a no-no place

    @scholrlea said in Mr. Jetson teaches network technology:

    you know, 1993 or so, before X11 had even been ported to Linux

    We were using X11 extensively with Linux in 1992. It was a heck of a lot better than the commercial Xservers available for DOS/Windows that we had access to. (Specifically, it could be tweaked to do 800×600 on genuine VGA hardware, though the monitors weren't very happy about it. Even though it was monochrome, the extra pixels made things hugely more usable; it was about 56% more pixels overall.) We started with the MCC Interim distribution (hard going!) and then switched to Slackware (much easier!) once that became available.

    And then people started getting hardware that was better than that… ;)



  • @dkf said in Mr. Jetson teaches network technology:

    @scholrlea said in Mr. Jetson teaches network technology:

    you know, 1993 or so, before X11 had even been ported to Linux

    We were using X11 extensively with Linux in 1992. It was a heck of a lot better than the commercial Xservers available for DOS/Windows that we had access to. (Specifically, it could be tweaked to do 800×600 on genuine VGA hardware, though the monitors weren't very happy about it. Even though it was monochrome, the extra pixels made things hugely more usable; it was about 56% more pixels overall.) We started with the MCC Interim distribution (hard going!) and then switched to Slackware (much easier!) once that became available.

    And then people started getting hardware that was better than that… ;)

    Ah, OK, things must have moved a lot faster than I thought they had; I mean, in Torvalds' original August 1991 announcement, he hadn't ported more than a handful of basic GNU utilities (I'm pretty sure it wasn't even self-hosted until just before he posted that, he specifically mentions that he'd ported GCC and bash and not much else), and I would have expected it would take more than a year to port everything needed to support X11, never mind X itself.

    TIL, I guess... unless you're memory is off and you were using 386BSD instead (which I suspect might be the case, as in 1992 Jolix was way better known, though IDN if X was ported to it or not), or Minix (though this was well before Minix 3, and I really doubt that X would or even could run on a real mode Unix-like), or one of the commercial Unices like SCO or Coherent. Or were using a workstation rather than a PC, which seems even more likely. Or this was actually several years later than you remember it being.

    However, since you specifically mentioned MCC Interrim Distribution (which apparently was really only in 1992-93, predating even Yggsdrasil by almost a year according to Wicked-Pedo), I am inclined to think you are right on this.

    So, unless you say otherwise, I will take your word for it. My own recollection is that three years later, in 1995, Slackware had a partial X server but not a really complete one (and no desktop to speak of - I think I tried to run bare-ass Motif at one point on it, and there was some kind of desktop available but it never worked right for me - though you wouldn't have needed one for the use-case you mentioned), but most of the graphical software that came with the distro CD was based on SDL instead. I suspect that my own memory s fuzzy, however.


  • Discourse touched me in a no-no place

    @scholrlea said in Mr. Jetson teaches network technology:

    TIL, I guess... unless you're memory is off and you were using 386BSD instead

    I had a friend who used that, but the machines we were running this on were a little bit short of space (the 16MB of HDD wasn't very much even then, and that was especially true once you took 4MB for swap) so the fact that we could use Linux and strip it down very hard was a big bonus. Also, I believe that 386BSD distributions assumed that it was running on a 386-DX whereas these machines were the 386-SX variant (without math coprocessor) and so the built-in kernel-level FP fallback behaviour was a real boon. And just getting the networking working (and not interfering with boot!) on those crappy old systems was a big enough challenge by itself.

    This all involves remembering details that I was hoping I'd forgotten. ;)

    It was enough space to run an Xserver with a full set of fonts that would then act as a front end to the SunOS (and early Solaris) machines in the lab, and that was great as it meant we could work from college after the bar closed. The code quality wasn't perhaps the best, but we could actually get stuff done near the Ballmer peak… and I'm pretty sure of the timing since I can remember it against which semester I was in of my undergraduate degree (which has very definite dates).



  • @dkf I am pretty sure it was the 486DX that was the first CPU/FPU combo chip. The difference between the 386DX and 386SX was the chopped address bus on the 386SX. The 386DX, which was originally just the 80386, had a true 32-bit bus, but since that was expensive at the time - not to mention problematic, given that the second round of PC bus wars was on-going and using a 32-bit bus meant not only having 32-bit memory units (a real headache when most memory subsystems still used a raftload of individual DIPs taking up most of the motherboard, as was often the case as late as early 1990) but also either licensing MCA, jumping on the EISA bandwagon, or sticking to ISA and sucking it up.

    As a stopgap solution, Intel came out with the 386SX, which was to the 386 what the 8088 was to the 8086: it was a 32-bit CPU but used a 16-bit external memory bus. It was a hack, and they ran much slower than the full 386DX did at the same clock speeds and memory wait states (remember when memory ran at something close to CPU speeds?), but a motherboard for one cost about half as much.

    The introduction of the 386SX basically carried the x86 PC clone market from early 1989 to the end of 1992, until prices for 32-bit hardware dropped (and the first wave of PCI motherboards hit the market), though you have to recall also that until mid-1990 or so, the majority of new PC systems were still cheap-ass XT class boxes (I should know, I was assembling PCs at the time and they were about a third of all the machines we built).

    (Those dinky real-mode boxes only really dropped off the radar after Windows 3.1 came out, and even then the main thing that drove them out was the rise of 2nd-gen PC gaming using DOS extenders and accelerated graphics cards - plenty of business users were happy with DOS boxes running WordPerfect and Lotus 1-2-3 right up to the release of Windows 95, and only switched then because the software vendors were leaning on it heavily by then and you couldn't get updated DOS versions any more. Hell, the loudest complaints about Win95 were really about having to learn how to use something other than WordPerfect.)

    Anyway, the point is that prior to the 486DX, the x87 co-processor was always a separate chip, and it was quite expensive to have it - hardly any PCs had one.

    Also, the main reason the 486SX existed at all was because in 1991 Intel were still shaking out the die process for the 486 - the 486SX wasn't one without the co-processor, it was one where the CPU worked but the FPU didn't, so they took all the flawed chip batches where the CPU still ran, bypassed the non-functional components in the chip housing circuitry, and re-badged them for a less expensive market (sort of like how in each new process the Celeron/Pentium/i3/i5/i7 are all the same chips, but sold as different ones because of variations in the production quality leading to differences in performance and functionality - which then means for later production runs, they often have to lock of parts of the chips sold to fill orders for the lower end, after the production issues have been debugged). This is an nearly-universal practice in chip fabbing going back to the 1960s, and IIRC was one of the innovations that Fairchild had left on the table, leading Noyce, Moore and Faggin to gather together a second Traitorous Eight and form Intel in the first place.

    (Ok. I am not being fair here, the split between Fairchild and Intel was a lot more amicable than between Shockley and Fairchild. FS even fronted most of the seed capital for Intel. IIRC, it happened mainly because Fairchild mgt felt that trying to manage both the production of their existing Small-Scale Integration products and the development of the Medium-Scale Integration memory chips Moore had devised would be too problematic, so they spun that project off.)


  • :belt_onion:

    @ben_lubar said in Mr. Jetson teaches network technology:

    @atazhaia said in Mr. Jetson teaches network technology:

    There was also no rule demanding that users must change passwords regularly (error nr 2).

    From the NIST:

    0_1518659288016_Screenshot 2018-02-14 at 19.47.52.png

    To be fair, that's a recent update. It previously encouraged periodic changes IIRC

    Also, that reminds me. I should send someone a nastygram asking why we're violating NIST best practices on passwords for no good reason...



  • @scholrlea You're almost right:

    • 486DX / 486SX: yes, with the caveat that some machines that shipped with a 486SX had a second socket for fitting a 487. That was really just a 486DX with a different pinout, and disabled the 486SX (but you had to have the SX fitted in order for the 487 to work).
    • 386DX / 386SX: the DX had 32-bit address and data buses, while the SX had, like the 286 before them, a 24-bit address bus and 16-bit data bus.
    • The 386DX could use either a 387DX FPU OR a 287, but didn't perform so well with a 287.
    • Cyrix complicated matters by making their 486DLC which was a 486-like architecture (without internal FPU) in a 386DX-compatible pinout. That had some oddities, since 386DX motherboards didn't really "get" the idea of a CPU with cache inside.


  • @atazhaia said in Mr. Jetson teaches network technology:

    @steve_the_cynic To his defense, he and I do agree on one point: that virtualisation is an important topic that needs to be brought up and properly taught, even if it's not in the course goals. And as long as it's Windows Server with a GUI he can give correct information and instructions for how to use it.

    I'd put money on him not knowing when the first hypervisor was available for actual customers to use. It's longer ago than most people realise.

    You know, like, um, 1967. (CP/CMS on the IBM System/360 Model 67.)



  • @dcon said in Mr. Jetson teaches network technology:

    @steve_the_cynic said in Mr. Jetson teaches network technology:

    @blakeyrat said in Mr. Jetson teaches network technology:

    @atazhaia said in Mr. Jetson teaches network technology:

    To connect your own network directly to the internet you need to apply with InterNIC to be granted one or several IP addresses.

    Correct, but incomplete-- strangely omits that you can just rent an IP address from someone else who has one. (I'm assuming in 2012, IPv6 wasn't popular enough to be worth talking about.)

    The last part is hard to say, but I'll just note that for the last year and a bit (i.e. since 23 December 2016), I have had a /56 IPv6 prefix at home, and therefore I have been renting 4722366482869645213696 IPv6 addresses. I think that's enough for all the computers I have.

    For now. IoT will take care of that.

    Ugh. Not if I can help it. It's bad enough that I have three "real" computers that use it, as well as a UTM-style firewall, several VMs, a pocket computer and a wrist computer. (Aka an iPhone and an Apple Watch.)

    And for all I know, the printer uses it as well.

    That lot will suffice for now, thanks.



  • @thegoryone said in Mr. Jetson teaches network technology:

    @atazhaia said in Mr. Jetson teaches network technology:

    This is done for example in MAC OSX which is built upon a version of FreeBSD Linux.

    Isn't that correct, at least, if very (very) badly explained (and Unix)?

    You think that "FreeBSD Linux" is in any way correct?

    And it's a very wrong statement anyway. macOS (formerly Mac OS X) is based on the NeXTSTEP kernel that in turn was derived from the Carnegie-Mellon Mach kernel, with some BSD userland stuff as well. No Linux involved, and not FreeBSD either.


  • Garbage Person

    @sloosecannon While NIST has seen the light, there are many other compliance standards that still require password rotation.


  • Winner of the 2016 Presidential Election

    @steve_the_cynic said in Mr. Jetson teaches network technology:

    @steve_the_cynic said in Mr. Jetson teaches network technology:

    @cursorkeys said in Mr. Jetson teaches network technology:

    @steve_the_cynic said in Mr. Jetson teaches network technology:

    Windows 10 won't boot off a GPT (bigger than 2TB) disk

    Definitely a load of crap, it's quite happy with 4TB GPT disks from experience. I suspect it was a non-UEFI motherboard and the disk was actually MBR, hence the 2TB limit which is a thing for MBR not GPT.

    I guess it might be just a case of them knowing how to set up disks with MBR, but not with GPT. And my point about preferring that they deliver something that works still stands.

    Well, the mighty Goggle pointed me at a handy, concise, and CLEAR page on Intel's website that explains how to persuade Windows 10 CU or later to switch its infrastructure from MBR to GPT without nuking the disk and reinstalling. That makes me happy.

    The most important part though is if it's correct.


  • Notification Spam Recipient

    @sloosecannon said in Mr. Jetson teaches network technology:

    To be fair, that's a recent update. It previously encouraged periodic changes IIRC

    As far back as 2016 it's always said that.

    0_1518850507587_9ec265e2-4db6-4c52-afd4-d681fa608c8e-image.png


  • Discourse touched me in a no-no place

    @steve_the_cynic said in Mr. Jetson teaches network technology:

    386DX / 386SX: the DX had 32-bit address and data buses, while the SX had, like the 286 before them, a 24-bit address bus and 16-bit data bus.

    But despite the differences you could use the same software with either. Unless you had more memory than the 24-bit address bus could talk about (and no, those horrible machines I was using didn't have 16MB, just 1MB).


  • Discourse touched me in a no-no place

    @tsaukpaetra said in Mr. Jetson teaches network technology:

    As far back as 2016 it's always said that.

    I remember when that came out it was reckoned to be a big deal that they didn't recommend rotating passwords periodically, as that had been something floating around in industry folklore as a “best” practice for decades. I sometimes think it started out like this: “I recommend changing the post-it you keep the password on because otherwise it ends up looking messy from dust and sunlight. About once a month will keep that piece of paper nice.” :headdesk:



  • @dreikin said in Mr. Jetson teaches network technology:

    @steve_the_cynic said in Mr. Jetson teaches network technology:

    @steve_the_cynic said in Mr. Jetson teaches network technology:

    @cursorkeys said in Mr. Jetson teaches network technology:

    @steve_the_cynic said in Mr. Jetson teaches network technology:

    Windows 10 won't boot off a GPT (bigger than 2TB) disk

    Definitely a load of crap, it's quite happy with 4TB GPT disks from experience. I suspect it was a non-UEFI motherboard and the disk was actually MBR, hence the 2TB limit which is a thing for MBR not GPT.

    I guess it might be just a case of them knowing how to set up disks with MBR, but not with GPT. And my point about preferring that they deliver something that works still stands.

    Well, the mighty Goggle pointed me at a handy, concise, and CLEAR page on Intel's website that explains how to persuade Windows 10 CU or later to switch its infrastructure from MBR to GPT without nuking the disk and reinstalling. That makes me happy.

    The most important part though is if it's correct.

    You're not wrong, but if it's not clear, it's impossible (or very difficult) to tell if it's correct.

    I did what the instructions said, and it did, indeed, just work. My system disk is now running in GPT.



  • @dkf said in Mr. Jetson teaches network technology:

    @steve_the_cynic said in Mr. Jetson teaches network technology:

    386DX / 386SX: the DX had 32-bit address and data buses, while the SX had, like the 286 before them, a 24-bit address bus and 16-bit data bus.

    But despite the differences you could use the same software with either. Unless you had more memory than the 24-bit address bus could talk about (and no, those horrible machines I was using didn't have 16MB, just 1MB).

    Correct, just like a 68008 would, subject to memory limitations, run any software written for a 68000. (There were behaviour differences on certain operations between the 68000/8 and the 68010 and above. The instructions to manipulate the full condition-code/modeflag register became supervisor-only in read and write.)



  • @steve_the_cynic said in Mr. Jetson teaches network technology:

    just like a 68008 would, subject to memory limitations, run any software written for a 68000.

    99.999999999% is not "any" - :) :) A certain real-time data acquisition system suffered timing issues when they released a low cost controller [68008]. The problems were subtle, and drove the engineers crazy as to what the root cause was [they focused on the system under test rather than the test system for weeks!]


  • Discourse touched me in a no-no place

    @steve_the_cynic said in Mr. Jetson teaches network technology:

    Correct, just like a 68008 would, subject to memory limitations, run any software written for a 68000.

    But the relationship between the 386 SX and DX wasn't the same. Not that it mattered much; the more than 16MB of RAM required to make a real difference was an unaffordable luxury…



  • @dkf said in Mr. Jetson teaches network technology:

    @steve_the_cynic said in Mr. Jetson teaches network technology:

    Correct, just like a 68008 would, subject to memory limitations, run any software written for a 68000.

    But the relationship between the 386 SX and DX wasn't the same. Not that it mattered much; the more than 16MB of RAM required to make a real difference was an unaffordable luxury…

    68000: 24 bit address bus, 16 bit data bus
    68008: 20 bit address bus, 8 bit data bus

    I guess it's not exactly the same, but it's the same concept: fewer address bits and a narrower data bus.



  • @thecpuwizard said in Mr. Jetson teaches network technology:

    @steve_the_cynic said in Mr. Jetson teaches network technology:

    just like a 68008 would, subject to memory limitations, run any software written for a 68000.

    99.999999999% is not "any" - :) :) A certain real-time data acquisition system suffered timing issues when they released a low cost controller [68008]. The problems were subtle, and drove the engineers crazy as to what the root cause was [they focused on the system under test rather than the test system for weeks!]

    It ran, but was affected by timing issues, suggesting that a 68000 was only barely fast enough, and the slower memory accesses (two complete cycles per 16-bit access instead of just one, etc.) were enough to tip it over. There weren't, to my knowledge, any semantic differences between the two processors, unlike the 68000/8-to-68010 transition.



  • @steve_the_cynic said in Mr. Jetson teaches network technology:

    It ran, but was affected by timing issues

    I am not sure I would use the word "ran", but yes, it was a timing issue. The killer was that all of the diagnostics passed and even the manufacturer was not aware of the problems.


  • Java Dev

    draws a pentagram, puts down some candles and opens the Necronomicon and recites some forbidden verses

    Arise, Mr. Jetson! Arise one final time to do my bidding until I banish you to the Forbidden Realm and burn your books upon the sacred fire!

    ominous rumbling happens and the unholy figure reappears from the ground, spouting insane gibberish and reciting network knowledge from millenias past

    So, there, with that out of the way, I am once again returning to the world of Mr. Jetson before I trash his books for good. I have noticed new and exciting things this far in my final review. Today's keywords are Copy&Paste:

    Upon reviewing a couple of books for different courses I noticed that about 80% of the books were exactly the same. The chapters where just copied straight off, just in a slightly different order due to one book having a couple extra chapters in the middle and the other one having them at the end. But yeah, lazy as fuck. Especially as reading the mandatory course content list, which is completely different between the courses. But nope, that wont stop Mr. Jetson from just hacking this together and calling it a day, and then wanting full price for both books.

    The other thing I noticed is that he seems to have lifted a picture off Wikipedia. Looking at the picture on Wikipedia it has a CC BY-SA 2.5 license. Does this book even attempt to even add any sort of attribution? Nope! Just a big fat © Jetson Textbooks plastered on every page. Naughty, naughty.


  • Java Dev

    Something slightly suggestive~

    DSC_0112.JPG

    Also notice how the formula in the illustration and the formula below the illustration are different. No explanation is given why.


  • Java Dev

    @Atazhaia said in Mr. Jetson teaches network technology:

    Also notice how the formula in the illustration and the formula below the illustration are different. No explanation is given why.

    E=hv and F=hf?



  • @Atazhaia said in Mr. Jetson teaches network technology:

    Something slightly suggestive~

    DSC_0112.JPG

    Also notice how the formula in the illustration and the formula below the illustration are different. No explanation is given why.

    They're the same, but they use different symbol<=>meaning conventions. One calls the frequency "f", while the other calls it ν ("nu" = Greek letter "n"), which is a different convention for representing the frequency in a physics equation. (Normally, in particle physics, I'd expect to see ν for frequency, not f.)


  • Java Dev

    @Steve_The_Cynic I suspected they could have the same meaning, as that happens a lot because people can never agree on one standard. Just a bit silly to not stick to the same representation in both places.



  • @Atazhaia said in Mr. Jetson teaches network technology:

    @Steve_The_Cynic I suspected they could have the same meaning, as that happens a lot because people can never agree on one standard. Just a bit silly to not stick to the same representation in both places.

    My guess--the author stole the graphic from someone else who used the ν convention and didn't bother changing it.

    Also--the situation you describe is totally, 100% normal (at least in US higher education). Textbook companies and writers are the absolute scum of the earth. A more wretched group of money-grubbing, stereotypical "greedy capitalists who don't care about anything but the next dollar" I have never met. We had a physics book that came out with a new edition. We ran a diff on the contents. The only change? Some of the problems were in different orders between editions. Same problems, but just enough of a change so that you'd have to buy the new edition and the old edition would become much less valuable.

    Working at a small independent school, the big publishers don't care about us. Their big buyers are the state departments of education who are ordering books for the entire state. Ordering 60-70? Not worth their time. Even if it's pure profit for them.


Log in to reply