 We are live welcome to block. There's a number 249 Wow cool, that's neat that's a lot of them and we made a 3d pumpkin post this on Twitter and this thing's pretty cool I You know in the spirit of Halloween, which is this weekend here nice great holiday lots of fun And this is just oddly satisfying to look at look at this just It's like hypnotizing And this is all 3d printed all these little parts here There's a tiny bit of glue that was used to hold this in but other than that It's actually just built and snapped together the screw that screw the spring that's kind of mesmerizing here actually will stay in there and Yeah, it's pretty cool. So this is definitely a little fun toy I made for those of you wondering if you look at my Twitter. I've got the link posted where you can download the 3d prints only a few tweets ago. So By coincidence we had some googly eyes because googly eyes are cool and this allowed us to You know make it easy to throw something in there. Where's my commenters? This is unusual that we have a live show with concurrent viewers and oh here we go now We got some people commenting It usually a few people start saying hello. I didn't I usually do this later and Yes, let's see. I just want to say thank you for your tips on pfSense and Neckate some of those videos actually us do this. We'll throw these on the screen as well Just want to say thank you for pfSense and tips and pfSense on that some of these videos of view Been have been very useful. My network in home is almost complete awesome. Hey, that's great to hear I am glad that is helpful. I Scheduled live stream a little bit earlier and I do it's just really nice. How I'm just not gonna lie I had a busy day and it's nice out So I don't plan on staying really late on the live stream today really pretty simple I had done a thing with Ninja RMM earlier. So the toe Satan is still sitting here Yes, we did some really spicy stuff at the end and it was just Yeah Ah It was fun. It was fun. It was spicy. It was crazy spicy now I figured today. We'll talk a little bit about me open up my lab So I can pull something up here actually and I got it done share it So click off the share share screen chrome tab There we go Well, it's kind of neat. I can look at this camera and see the output of this Cool, but I want to talk a little bit about, you know, while we hear we'll talk a little bit about benchmarking Lies damn lies and benchmarks as I've said before when I've done some of these type of benchmarking videos But a few people have always asked me how you're doing a benchmark So I always say pharaonic test week, but I wanted to kind of Talk about it a little bit and I'm wondering if there's a I don't because I don't really I didn't really look I mostly just did reading But I don't think there's really many people who've done videos about using pharaonic test week France test week is actually really cool and it's it's really in depth How well it works actually pull that up too So let me just stop Share Pull that up there we go and One of the things abouts pharaonics and why I like it so much is one the library of of The just the crazy library of all the tests you can run is really outstanding with pharaonics It gives you just an absolute ton of options, and I really like that. It also is good for Scheduling tests that are rather complex You can actually set up the pharaonics server and it's like a web interface where you can have multiple machines Schedule all the tests send all the jobs make them run concurrently or on some type of schedule Like you like them to run and it gives you automatic uploads right to the open benchmarks Now having a common benchmark that I can run across machines and someone else can run across machines So we can have a consensus of this version of this benchmark run with these parameters So we can understand and test things is huge to market so we can have like a consistency Of how we look at things. It's one of those Challenges when someone says how fast is it and fast is a real relative term Because so much of it comes down to what's useful to you What do you need faster transfer speeds faster IOPS and storage? It is the One neck eight has their listings It's one of the reasons they don't just list like the IPERF traffic because IPERF is well singular streaming not complex when you look at a Network it's going to be a series of complex streams and how well do those go across? Well, these are all those things like you have to have common languages and common tools To reach a consensus. So we both are talking about the same thing So state the goal and then have benchmarks that run till we can achieve Similar results that work for the workloads you're targeting. So it's kind of funny It's kind of fun playing with Pheronics and you know doing all these benchmarking tests that I've been doing because it just brings me closer to that consensus And it's numbers that you can repeat yourself if you want to know Hey, how does this system can figured this way perform and then you'll get an idea that you're like Oh, I need a workload that would actually be able to do that Or at least achieve those performance numbers or a minimum for my type of workload because it's this and uh, yeah It's just kind of a fun thing to play with uh Been using unify access points combo in a hospital where I'm admin best choice ever We've we've done some big medical facilities with them and they work fine We haven't really had issues with them. I mean they It's one of those things like when people ask me about unify Someone may point out that there's more features in another brand. Do you need those features? Are you going to use them? That that's what it comes down to I mean if you need those other features that aren't available in any is goes for any product If the product has the features you want and you're not going to use those extra features that the competing product has Then comparing it to those features never made sense to me because I'll see people But this one does this other thing But do you need that or think well not at all? So why do you just because a product can do more doesn't make it the better fit for that particular project? We got All the way from Malaysia. Hello from Malaysia catching on live. So doing it earlier. I'll probably get a few different faces here So that's great a few different people. Um, let me find Clear what I have open Drag over another screen. So it's closer to me. Stop sharing And then we want to share a screen a window I think that's the right window Yeah, there we go Now I got to go in and uh So we can log into feronics. I figure we'll do a little bit playing with it here Let's make the screen a little bigger All right, there we go And we'll just run this test real quick Well, there's a few different tests we can run but the feronics test suite And like I said, maybe there's a demand that I'm not aware of of making a whole video on this right here But the feronics test suite a matter of fact, we can probably just do this to get a whole I think there's a list suite in here Analyze run times build suite list all tests And we can listen all the different tests they have And uh, you know, there's another one here is a sequel test. So we can run the sequel light speed test That's specifically for a system. So we'll go ahead and do this So it would be feronics test suite And it downloads it. It's gonna run this real quick Actually, let me do something so I can see the comments over here too All right, it's downloading that Keep an eye on the comments here Hey from south gate. It's uh, Brett shit. I'm there. He's in the other office. He's not just in south gate He's like in the other side of the room Bulgaria is not Bulgaria is much further away While that's downloading in your udm pro review, you mentioned that the hardware was great But I have some software limitations Multi ip in a single point when have those other limitations been addressed? No I did a recent video on the udm like very recently about why you I mean What I talked about what works and what doesn't so I my video was like maybe from two weeks ago Maybe three and I don't and people keep asking this question I don't understand Like I keep repeating myself on this one and I said it in the video The the limitations are so much software based So and someone says well, what about the new hardware they're coming out? Well, if that new hardware isn't gonna accompany a update to the software to fix the shortcomings It's the same thing again with the same problems. This is the problem people have they go Oh, the usg wasn't that great. So let me try the usg pro and they go Hey, the usg pro has all these same software limitations. Sure does This is one of the problems with the way unify just is not putting the time they really need into Improving the product to eat the next level for firewalls It's a shame they don't because it would be a more popular product if they did But it they didn't so here we are The other day you showed a typical setup which you guys have for a client Which is the psn router and unify switch But you can't run a unify switch without adopting it via unify controller. That's true We if if we're going to have a managed client with unify switches, we manage those clients under our controller If you want a not managed if you want a switch that has a web interface, you just don't get a unify I mean, we've got clients that are one-off clients. They're not actively managed that we install standard switches in You know, we'll we've used sysco switches. We've used because I've used some of the sysco catalyst 1000 switches We've used some of the uh edge switches kind of depends on the client We talked to a client recently It's nothing we're doing active engagement with that has some back gear switches They know how to configure them. They have them set up. They have a web interface I mean if they're functioning and there's not a reason to rip them out use them But yeah, if you use the unify it's got to be adopted to a controller that controller doesn't need to be on all the time by the way So let's kick off a benchmark in the background while this is going. So let's go ahead and uh, we'll do Threads you can do more than one test by putting commas, but we'll kick this off so it runs in the background We'll say we want number three Would you like to save these rolls? We'll call it vlog Thursday a unique name Test run for vlog All right, it says it's going to take about 33 minutes. See how long it gets the estimate wrong all the time About how long this will actually take to do. So we'll go ahead and keep answering comments while that runs in the background Ah Yes, you can run the controller a glenic cloud key I don't really recommend in windows But we do have some people because they only have a windows server They only fire up the controller to make changes. Uh, and then they shut the controller back down I mean that that even works. So That's uh, one of the options on there Actually, we should probably switch. Let's see. Is there anything interesting to watch here? Yes We'll pull this up now while that test runs. Wow, that's not bad at all We share the results with you You know, I wonder if I can pull them up on the others Can you share two screens at a time? That's that's the real question if I did a share screen And I did a Actually cancel. I've not tried this in stream yet. We're in new and uncharted territories here What happens when we go here and hit share Share screen new tasks. Let's see what happens Nope It doesn't let me uh Okay, I have to share it from my other computer fair enough fair enough So We will remove this one and I'll add back the other one Actually, let's do this. We're going to Share Stop screen share share screen. We'll just share the whole windows like this drag things around Entire screen screen two share We get this fun effect again. Now we can do this Then I can tab over to this But this is what I see on the other side so we can try to figure out how many iops It is now. This is looking at here. We can also log into this analogy itself maybe Thought it was a r s three There we go Because when you're running benchmarks, you got to log into all the things you're benchmarking Pull up the resource monitor. See what it looks like on here Let that run let this run Pull this up. Oh, okay This one it's actually already done here. So not much to look at now This is how for onyx looks when it's done. We're going to get it wants to upload it to open benchmarks. Yes Yes, don't share all my data Happy link And we can see the test results So here's the test run for the sequel light test and what we were able to get now Let's go ahead and test something here. This is you know, these are all the steps you go through when you're testing Let's go over here to the disc. Currently this disc is living on a Synology rs 36 21 if you didn't see I just released a video a little while ago and it was my testing on that We're going to live migrate this To another search. Let's do the local storage So this will be a live migration We're just going to pop the hard drive over to another storage and we're going to run under test and see what happens so it takes about two minutes to migrate the vm to somewhere else the storage for the vm specifically and See how that goes Let's go ahead and I like when it looks like this. Let me go ahead and uh How can we fix that? Oh, I know how can do this Let's cover this part up This is how I see all the chat. This is still doubled up. So we'll go ahead and just remove it for now and answer questions Oh, let's see Hey, tom. I was in the u.s. I bought some unify ap's back in my country. I got a warning message. You current country of chantry is not supported Uh, is it risky to live like this from now on? There are rules that you may be violating. Uh, I don't know what the consequences of breaking those rules are but you should probably look that up uh, the Wi-Fi ranges the any type of broadcast is regulated in each country by their designated things and here we have the um fcc that manages the airwaves here the bandwidth is controlled by the government And they have restrictions on it. And if you're in another jurisdiction, they may have different restrictions So that's why they have these rules like that. Uh, I don't recommend breaking those rules. I don't know all the consequences, but um Oh awesome Lawlessness and wi-fi in your country are no rules that I don't know. I don't uh You're getting a warning because you're outside of the area that that is known to be allowed. So Oh, yeah migration of storage of live machines. You can live migrate things wherever you want and it it happens really fast It's like already done. Um, so our storage is now migrated to a different machine So let's go ahead and uh, what screen should we share again? Pull up this one See this is all migrated. So if we go here and look it's now on local storage And by the way, we didn't have to reboot it. But what we can do now is this We'll run the same test again I think we chose option two Uh, actually I can you can look at the results real quick 32 all right. So that is option three Hold on to me flipping screens fast here folks When you do it all yourself, it's uh, it's a lot more so option three Would you like to say these old sure would what we like to call it? We're going to call it vlog thursday again Because we wanted to upload to the same spot. So we'll say vlog thursday and uh, we'll call us when using local storage All right Now it's going to do this test run and now we're going to find out his local storage on this particular setup faster Than the storage on the other one. So let's go back over But this is kind of the neat way it works. So you can go through these benchmarks now when you keep using the same benchmark name over and over again And add a different description. It consolidates them all together to do comparisons So now let me pull this up. So I see when it's done once this Benchmark is done We're going to see the comparison between them and I'll be able to judge which one's faster local or The synology storage local storage is mvme. I'm going to guess the mvme is going to be a little bit faster I say a little bit because the synology if you didn't watch that our video is a 12 drive all ssd system So it is really fast with uh 25 gig connection in it. So yes, it can definitely uh handle some speed As a matter of fact, we should probably after this is done. We'll set up another uh Another test. I have some window servers We can copy over to it as well and run like crystal disk info if you want to see how fast the synology is versus the local storage as Well, but these are all those benchmarks that we keep running to keep doing these tests I say windows the windows is we're going to run crystal disk not pharaonics I think there's a way to get pharaonics working in windows. I always run it in linux Because most of the workloads that outside of windows other than Knowing what you need for windows workload most of the stuff that's going to run in these services is a bunch of linux stuff It all depends it varies on obviously what the client requests are and things like that Still running You can bring that back up You know, I could probably drag this window over here I need like one more screen maybe Because this one makes has got me rubber necking because I look way over to see the see my other screen over there But this would allow me to Leave this up and uh, let's see if that's done switch to this over here. That's still running Yeah, leave that pulled up United States seems to be the only country where wi-fi hardware has to be physically limited where other countries allow you to choose Um, I probably would guess that's just because we have too many things here Uh, that would overlap. So they they narrow the bandwidth pretty much for a lot of restrictions There's just a lot of things going on here a lot of industry. So they if they didn't do this, uh You would end up with this giant overlap people Broadcasting things they shouldn't and some of the problems that may come from that actually I gotta turn this other camera off There not even using that camera. I don't need that one on That'll be kind of weird. I could there's a few different camera angles I can switch to but that would be complicated right now. So Um What type of remote storage it is an eye scuzzy attached to synology So that's a good question. So just so you can see we go to storage We have this synology and then we go here to the eye scuzzy manager and Here's that we can also open up the storage manager And you can see that we have 12 ssds in there. Um in this particular system. So it's eye scuzzy connected to xcp ng Um 10 gig to a synology with 10 gig and ssds is enough for you Yeah, you know until you get a 25 gig and then you're like, yeah, I definitely could go more Yeah, I usually can I I have churnass servers that I do testing on too. Matter of fact, I have some new churnass benchmarks I'll be doing uh soon I've done a video on uh Aruba instant on before uh the ruben sun's gonna be The aruba instant on is going to be The um What do you call it? The easier platform to use and has a dashboard to manage it Uh, but they do have a lot of the instant on devices have local access web interfaces like on the switches So, I mean, it's kind of comes down to whatever choice Whatever choice you really want to Go with me go takes got a lot of features but comes at the expense of being really complicated So that's one thing I don't like about them Uh, if you send emails to me they usually go into bit bucket. So you need to post in my forums So I I get so many emails. I don't reply a lot that's I have Different email addresses I use that I can't share out Everything has to get filtered through either my forums or a ticket. Um, because the volume of people that just randomly email me is Too much sometimes and so There's that there's like 22 000 messages in my inbox sometimes. So how many are there now? Let's look how many Uh Yeah, so emails generally Don't always have time for all of them That's why I use the forums 242 000 overall emails and there's a percentage of those that are unread So if you if you had an idea of this Once you had an email just for a lot of people to send it and you just have to switch and create new email addresses Um, so this is an fyi on people wanting to just randomly send me emails It's not likely you get a reply Did you see Linus switch to true dance? Yes. Um, I'm happy to see that Linus I tweeted this that he's finally come on over to the enlightened side of how were you trying to run all of that on window storage server? Uh, it's just not practical And so that I surely these wires are dangling. They should probably be up here really long There we go. I didn't know if there any I don't even know if anyone knows certain screen or not I just looked over and like yeah, they're kind of hanging there So yes, Linus is now finally with the true nas If I had a nickel for every time someone said please submit a ticket Do you know even for us? It's when you fill out the form on our website? It goes through our ticketing system and then the team deals with it so to speak I mean, I'm among that team that looks at all the tickets Well, I can't look at all of them because it comes down to time And that's just something it's a scalability problem too. I don't have time to do one to one Unless people are buying time for me. I don't do one to one support On email. It's impractical. This is why I have my forums. The forums are hugely useful because people keep asking the same questions Which is great. That's fine. They're the same questions are common And the way the forums solve that one to many problem is you're asking a question that when you start typing in my forums You go, oh someone answered this already. I'm not the first person that had a dhcp problem I'm not the first person to have a dns problem or a misunderstanding how VLANs work. So this is one of the ways I solve it is Heading over to the forums a matter of fact one of my favorite things on the forums is I see people post there and before I have a chance to reply There's often people who have the answer for you and do reply So this has been this is why I have to push it that way of me answering all the emails Is just like I said, it's kind of an impractical thing Upload the test results. Yes Yes Whoops That isn't what I wanted. What did I do? Not like open a new window Oh crap My tmux session is broken. Oh, well I think I can get that test again Did I copy it though? Nope I can run it again Such as life. Whoops I accidentally killed the session. My bad Ah fun stuff Linus is finally growing the size. They're starting to see real world problems. Yeah I think he was seeing them before He was seeing the problems before it's just like Seem like he really opened up to non windows option in the past few years Well, it's one of those things if you've worked in large scale systems, you'll realize they're not run on windows Windows doesn't scale to the large scale cloud systems out there They're constantly re-engineering azure and things like that to try to make it work And uh, yeah, it's it's linux has been there for a while Especially when it comes to storage servers True nas has been in the enterprise market on the back end for a long time ZFS is not new. So that's definitely um It's just the reality of it All right, so we can let's run that test one more time So we'll say actually let's run this test Yes We'll call it the vlog. There's day one again. Maybe we won't close the window One more test. I think if I do this it'll actually upload all the results I could probably just force it to upload the results. But if we do a single test right now, it'll When this uh, what does local storage look like on this? No, not bad. Last time it was hitting about 60 000 IOPS on here So here we come ramping up again With a single test Is that true nas running as a vm on zen server? No, it is not. I am not running true nas as a vm I don't ever recommend running true nas as a vm outside of a lab environment Yes, someone's going to point out, but I got it to work. Yes And then there's more people that didn't get it to work If you got it to work hats off awesome But then some people I knew that had it working an update came a new kernel update and it broke again It's not true nas was designed directly to interact with hardware And even when you do full pass through there is occasional problems that you may run into It's not something I would run in production But if you want to run it and have at it, I've talked about it. I hit a video on it But I also say it's for a lab more so than anything else It's not something I really Would push for in production Yeah, linus has some Network engineers and linux people working there now. So yes, there's he's hired a lot of people Yeah, any known drawbacks running true nas as a vm many many drawbacks like random crashing corrupted drives Random crc errors. We've had people contact us for troubleshooting And we started digging into troubleshooting and right away realized they had it virtualized like We couldn't solve the problem now. There's a good write up on uh If you dig around for a virtualized true nas Someone found a kernel bug and the people at xcp and g helped modify They worked. I believe with the people at uh ix systems to help solve a problem and pushed an update to help Solve a condition that was occurring But that's the kind of level it takes it's some you can run into some obscure problems And one of them is uh me and Xavier worked on his system He virtualized it because he wanted to and he constantly had problems with it He constantly got a pool that worked fine for some length of time and then ran into random crc errors Not consistently, but if you load the system bare metal works fine So it was only occurring when he dealt with it when it was virtualized So the hardware was fine. It just didn't like true nas being virtualized even though we did all the proper things in pass through and uh, yeah It's Just one of those things that happens Now we can see the final results here There we go Oh, I need to run. Yeah, there's the True nas the first one local source. So actually local the local mvme is Oh fewer is better. So local local mvme is actually a little bit faster by like double so We can run more tests too and keep adding them in here. So it's another test we can run like the fi o tests one comma two Actually before we run it, uh, we want to migrate it. Let's put this back over. Well, no, so I guess we can run it here So we'll run this test again. So one comma two. We just want to do some random read writes for Yes, and this is going to call We'll call this vlog thursday local mvme All right, so this is the local mvme and then we'll move it back over the technology and run more tests This guy, you know, so we're just walking through how all this works for those of you that are interested in this I just got the best way to read more storage or is there better options at a set? So don't do it over smb. smb is a bad way to do uh vm storage So I did a video comparing ice guzzi and nfs if you're using a synology ice guzzi is noticeably faster, but um, there's some different things like if you're using xcp and g with ice guzzi It doesn't thin provision. So that's that's one of your disadvantages. But if you go over to True nas with nfs and ice guzzi the performance difference is much more narrow Ice guzzi still outperformed in true nas a little bit better But it's not like blow you away night and day double your performance and back to specifically to xcp and g It's not thin provisioned so that can be a disadvantage on there And I actually like I I set all my stuff up for my production equipment all in nfs There might be some edge cases where I've used ice guzzi but everything we have here is run over nfs bounce because it just It's easier to deal with because for when you're snapshotting every File on there every vhd is a singular file So it's easy if you ever have to run snapshots on there and on the true nas server on an nfs Share and then flip those snapshots back You can just grab a single file and flip it back So, you know, you can just run run it like that. It makes it a lot simpler and uh life is good Hey, thanks for the thoughts everyone. Cool. Uh, you're gonna go with the meekertik seems to be much more popular over the aruba um The thing is uses switch os. That's one thing about the meekertik I really I got a video on switch os. There's switch os. I actually kind of like it makes the system a lot more manageable Compared to using their full maker tick If you're space limited running shoe dance virtualize is good for backup storage. If you say so, I wouldn't trust it That's my my personal take is don't trust it for production Getting my homeland started of two r 720 xd's on the way awesome You need at least three for an ha cluster Just so you know and that's because of There we go Split brain computing look this up. Uh, this is why you need at least three Uh computers for ha so uh, it's called split brain computing You can dive into that topic right there But that's um, if you want to do a proper ha You'd want to look at that but it's so you can still cluster you can still Well pool them together as the way xcp and g uh is the nomenclature if you're if you're using xcp and g use whatever makes you happy um but yes SMB doesn't handle the file locking quite as efficiently. That's why smb is bad for vm storage nfs is still You're going to get better performance on it. It comes down to the way it handles certain file locking I don't think the smb Protocol deals well with all the frequent small writes with open file locks. I think it has a hard time with that Um, I just know from performance testing it doesn't perform all that well Speaking of performance not bad 725 This is our uh reads and these are our writes at 660 so not bad That's about uh, wow 191,000 I ops that's 168,000 I ops. I'm going to say the local search. This is just a standard rising system by the way. This is like a Uh, I I'm going to work on a video on this. This is like an off-the-shelf Rising from microcenter with a motherboard. This is all consumer stuff Um, we just wanted to build it for a test lab that had a rise on it That was inexpensive with an mvme. There's no redundancy on this mvme But hey, it's still impressive the kind of numbers you can get out of it So it's a fast system It was xc xcp and g which shows storage stats while one of the hosts are down um Hi, I mean the stats are It can't really show stats when a host is down because there's no stats to show those if you look at I mean, I get it. There's no stats. You're right. I guess you I guess you're looking for the historical stats, for example Um on this right here. I guess that's probably what you're looking for Yeah, it does not show the historical stats, but You can get an idea of them from here Which but then of course it's mixed in with everything else Yeah, I've never looked into microsoft solution for it. I don't think we're actually working with we've worked more than one Company that's just getting away from hyper v and all the problems. They've had with it. They found um The xcp and g to be very stable I was talking to wendell the other day from level one text. He's been really getting into a lot more xcp and g He's really impressed with the product as well like I am so xcp and g is where it's at man. It's a um It's pretty awesome. It's I'm gonna do I have a new video I'm gonna be doing on it because they added so many cool features in here for like their backups their smart backups and deltas they've really done a lot of enhancement in here of how all this works and It's been kind of fun to play with all this It's just benchmark done Yeah, still running I would love a lab rack with servers for some performance testing, but I don't I want all the noise in my condo. Yeah Yeah, it does It does uh have a noise expense that comes at rent a colo. That's another solution for that. Just run some colo space Yeah, there might be some way to do it. I I don't know how to there's actually a thing that i've never tried It's called um It's not something I've ever played with but it does claim to work with xcp and g This is this is like I got it's not my to-do list to try this because it's novel and it's been around for a while And they've updated it to you know have support It supports citrix. It supports xcp and g I don't know how fast it is with the way they're doing it though Because they have a slightly way they're synchronizing everything right between there So if you're keeping drive synced in order to give it the ha across these servers, it's doable But I think it comes probably at quite a performance penalty. That's that's why it needs some testing I don't know what that performance penalty is, you know, that's what that's what benchmarks are for so we can figure that out So if I set it up, I'll be benchmarking it. It's kind of a novel novel idea and it's been around for a little while so Uh tests are still running here Do you need to have a windows host to manage xcp and g? No Uh, I am clearly running. Oh, in case it's not clear. I'm running linux. I am obviously I'm managing xcp and g It's all web interface driven with zen orchestras. So no you you can do this I don't even like the windows uh management tool at all for it I only like using it for that In case you're wondering too the reason this test is running so long when it keeps seeing deviations It keeps reiterating the test until it creates an average until it gets a consistent benchmark Not the historical stats for a vm. I'm talking about the stats of share storage when I only have One of my share storage hosts off the stats go away Yeah, if they're I guess that's true, too. If they're down, I can't see them um, I never use it in docker I I imagine he could but it's so easy to build I have a couple. I have a debi and 10 build video for it. So I use it with debi and 10 and I build it um From source. So it's there's a script that builds it for you. Is this why I don't that's why I don't run it in docker Also, I don't know if the docker Um one is officially by them I don't like when docker containers are by third people third party people I don't know because I don't trust them if it's From the actual project maintainer great If it's somebody that just does it because they want to Less great unless I know them especially with all the supply chain attacks I mean if you haven't noticed there's been a giant increase On well, I covered this before there's been a bunch of bad docker images that had backdoors at them Because some people don't do things out of the kindness of their heart They do them to build popular projects slip in a back door Put popular project out there It's a it's a pretty solid solution because people so many times without even thinking about you're like just pull a docker image Docker pull the magic thing I want look I found it on a website It says docker pull and this container can be built and have everything I ever wanted in it and uh, yes yes Yes, yes So we have our first result here No, let's see Only the random read a random read random write all right Let's run that same test again. But first must migrate the storage and I love that. We don't even have to reboot this We're just going to go ahead and migrate Move it over to Synology and in a minute It'll be over to the Synology. It's on its way Send it over xcp and g just makes this testing at least somewhat easy You know not I mean not that rebooting the VMs as hard this machine's relatively fast So I go back over here to the Here we go. Here's here's all the data coming in Look at the network traffic I tried you tried h. A lizard Didn't like the disk base limitations found a better solution to be nfs on top of dr db. Okay Ah, yes No, gs certainly did suffer this so they're attacking certainly threat actors are attacking no, um It's actually I think it was the mpm libraries that got attacked that ended up elsewhere But there the build supply chain is under heavy attack trusting random people to build you docker images Just seems like a horrible idea to me. Uh, so I definitely avoid it I carefully choose my packages and even though I'm being careful. I've only mitigated some risk I've certainly not 100% unless I write every line of code myself mitigated risk because yeah Uh, it's a real challenge I don't use cloud init, but I think j is working on or has done a cloud init video. I don't use it Um, so I can't really answer any questions about it I know j's been working on some cloud init stuff. I can't recall if he has a video on it or not Learn linux tv, but you know what we can do learn linux tv Yes, he does have that one on there So j has a cloud init video and it's cloud init in general I haven't tried it specifically with any add-ons from xcpng But uh, learn linux tv does have a video on cloud init One of our challenges for 2020 making sure containers are sure Yes, two charts is definitely taking this head on and this is one of those things you deal with You have to have trusted sources where you get the container from It if that person who is maintaining it has bad intentions Those bad intentions carry on over to you and it's a real problem that people face So that's uh Yeah, it's it's a scary thought. Hey look, we're living on the Synology now So now we just uh do a little up arrow one comma two yes We're gonna call it Vlog Thursday A unique name. This is Synology All right, and it's we'll run this test right here And then we can compare the two is the local mvme faster or the Synology faster So that that's what our next thing is. So we'll let that run in the background here It'll it's ramping up doing its thing Then we'll see the Synology here do the overview Peaking out the network peeking out the ice guzzy 2022 hey 2020 2022 to future. We'll just say the future Yes Didn't the time just quit moving forward after 2020 isn't that really what we're dealing with here Since 20 it's just 2020 2020 hit and it's just after this it's 2020 It's just who even counts the years after this one Ooh 535 and 135,000 iops for this not bad lots of results Yeah, the dev of a library got his github account hacked and the hackers uploaded a new version With the crypto miners code that is correct uh Steve gibson covered this in security now broke down some of the details of it. You it's not hard to find You know these type of attacks and I did a video on dependency confusion This is a real problem. Like I said the supply chain attacks are ramping up and This dependency and confusion was done by a security researcher Which is so good because one wow was that an interesting write up but two it's also um It was a warning shot because a lot of the he was the person the security teacher that did the dependency confusion write up Was able to get a lot further than he should have in some very large companies by an attack a lot of people just weren't really looking for Um, so i'm glad a security researcher was doing it before someone else um Facebook in some way solved identity for social networks. I think facebook has done nothing in solving identity for social networks Nothing at all. I feel There's so many fake accounts on facebook People impersonating other people No, I the And most of the time it's not an identity crisis problem. It's people who still don't have 2fa under accounts problem That's usually where the issues start with a lot of these. Um, there's the developers just have to tighten it up a little bit Especially the older developers that were just used to doing it a certain way and go Oh, no one would really go after me. Would they I just make this one little Module that doesn't seem like that big of a deal even though it's used in 130 million projects So that's yes What research are you talking about? Uh, see This right here great read dependency confusion. How I hacked apple microsoft and dozens of other companies Yes, the story of a novel supply chain attack. This is a brilliant write-up and basically it turns out that They were able to figure this out and this is so clever that when these large companies, um I think I can drop a link to in here these large companies They were able to discover certain dependencies. They were requesting now the dependencies. They were requesting the namespaces Were internal, but it still made an external request. So let's say we're looking for Uh special facebook dot whatever or actually did you see facebook or is it apple was in there? Who's the companies on the list? apple microsoft and dozens of other companies So let's say apple's building one of their internal tools or microsoft or any of the other companies in this list here Know it has the npm name red sheet and python pip python package index package and rubies gems can be found as well as Really jump so what they're doing is they're attacking these namespaces So when you're building your tools, you have your external tools that you're pulling down and you have your internal ones But the search order happens to be in the build process This is why this is a novel attack and it's it's it's mitigated now because this was from a little while ago At least I hope it's mitigated. I really really hope it's mitigated But what happens is in the namespace The I have these list of dependencies I need to pull and some of them are internal dependencies that I have that we wrote internal code for You know i'm the developer I put these internal code and I have a code call for uh pumpkin 2.0 If you know that pumpkin 2.0 is needed it goes and pulls it but the system is also smart and wants to pull the latest version It reaches out externally and would pull things like Versions that are higher numbers. So if you know it's pulling pumpkin 2.0 and that dependencies met locally it pulls it But if you create an external dependency in the same namespace and you put it pumpkin 3.0 It goes I found a new version of pumpkin so it would start pulling code in and people are like Well, it shouldn't work because then that code wouldn't match because unless you have access to the internal code There would be no way to do that But that's where you're wrong this person breaks down exactly the couple edge cases where that can happen because normally I wouldn't know what their internal code looked like. So it would be really hard for me to match the code but then modify it That's where things get a little bit clever and a little over my pay breaks. I'm not a developer, but he breaks down how To work around that and how he was able to get it now. Yes, the product would probably fail the build But he's in that's the that was the important thing here So yes, it wouldn't build the software wouldn't work right because we injected a completely wrong version of pumpkin that didn't match the internal code base But you were able to get code to run and build in there that may execute something and then be inside the system So it's a giant mess. That's all there is to it That's the dependency confusion in a nutshell. I have a video where I think I broke that down as well Right talk about that 138,000 iops. This is pretty fast. So this is still running in the background We assume it's all mitigated. I'm sure we can sleep well at night. Yep. Yep. Yep Someone tried to personally me on facebook facebook only did something after a couple of my friends reported despite reporting them 15 minutes After they were discovered. Yep. It's slow for them to fix that problem It's that's what I said facebook's really done nothing in my opinion for identity management There's so many facebook accounts that are fake. I see it all the time. I especially um Less tech savvy people click on links all the time and people clone their facebook account and then start reaching out to the People they know and it's just it's messy Nothing good comes of it. Nothing good comes of it How much longer does this test have? Oh, it's getting better Matter of fact, we can see some of the results here Ooh 133,000 versus 186,000. So yeah, local MVME is certainly faster than the Synology there But yeah, that's life It's not bad. Don't get me wrong. 133,000 Um iops is a pretty good number, but 186,000 is a better number I like how it does little bar graphs here to give you an idea of how how much faster it's going Actually, we have however I think I'll let this test run and then I got to jump off because I'm going to go Do things outside because it's nice out. It's not supposed to be nice this weekend, but it's nice out today So I'm going to go be outside today Oh, let's see What else seems to be up here? Let's go back over to zen orchestra Not bad. Good speeds here More testing I'll get this benchmark updated Well, actually here we'll I'll throw Jay's link in here too um In case anyone's wanting to watch the getting started with cloud init by Jay, so Planning on replacing my udm pro with pf sense with the sg 2100 be a good replacement um Yeah, I think that I think the sg 2100 is what I use at home as long as you aren't Uh exceeding or or I'd say I think it does like 800 fine once you get over 800 meg for internet It's uh, it chokes a little bit. It can do it. I think I can't remember exactly where the limit is My internet at home is not 800. So it's it works fine. I only have like a 200 meg connection at home So for me, it's been working perfectly fine. But yeah, as long I think it's solid That's actually like I said, that's what I'm using right now at home is still a 2100 It's been my go-to for a while. It's it's a great system. Look at all this sonology testing on here Yeah, I wanted I'm going to be doing more videos on how we handle all of this All of our restores our backups. I gotta do a new tour of the lab just keeps getting bigger and bigger and uh These are pages of vms that we have here available to my lab Or these are actually just the logs the pages we have here. It's not that many pages Oh, we purged a lot of stuff. There's only two pages How we restore and back up everything They've done so much good engineering on this. I'm really excited about how they've done this with cpng I can't wait I turned off my udn pro. Hopefully the last time this week Yeah It's it works for people who have basic routing needs once you go over those basic routing needs. It's kind of You know, it just it's not a good fit Almost done. What are the other questions we have? What is the other q&a we want to do before I get off of here? Because this this will probably be done shortly and we'll look at the results and shrug our shoulders because the sonology results Are way more interesting here Drag them over Here's all the benchmarks. I ran on a sonology earlier It's a lot This is where I was actually hitting because this is um If you do the math here 1,721 megs is actually about close to 13 gigs. So that's impressive That's good numbers on there I am not you I don't really use unms. I'm not someone who can really help you much with that We're not much of a unms user Oh, yeah, I need to keep it around to retrieve video footage. Yep Yeah, 20 watts ain't bad a lot of the older i5 systems usually use more than that though I'd say 20 watts is good most of them Overall not just the processor, but the overall usage is uh more than that That's definitely um, definitely an issue Still running Almost done I don't know. Is there is there a need leave me a comment here Is there any need for me to do a more in-depth video on foronix? I think a lot of people that use it kind of get it. So I don't know if there's really um How much of a need there is for it Oh, uh, curious what those sounds you don't be with raid. So if We actually ran into a problem and I talked about this in my analogy. The analogy video is the video just before this one um Please note raid zero on 12 drives I I hit the hardware limitation So that's actually one of the problems I ran into is These benchmarks should be different. They're not they're they're a hardware limitation It's 13 gigabit seems to be as much data as I could pull out of here So raid zero should have really been faster But it wasn't because it was a hardware limitation. So that's Sometimes you run into that because you're running solid states The limitation isn't the raid type or the hard drives anymore The limitation is the fact that this is as fast as much data that will come out of that system and attach to the next system Hey, awesome. Glad I'm helpful Speaking of astrology, what are one of the best one for a home user who has files and picks They want to back up instead of paying for cloud services Really, whichever one's big enough to hold however many photos you have and how many photos you plan to grow into This is where the concern really comes in is is it big enough for your storage needs? Are your storage needs the same as mine? I don't know. I have seventy Eight thousand photos. Um, that's a different storage need than someone who has seven thousand photos So it really comes down to which system are you going to get? Is how much storage do you need and you know? If you're someone who's more casual and less obsessive than me with years and years of stuff that you Just hold on to I I am very much a minimalist in my life, but I am a digital hoarder So I I keep all my digital things. I have them all organized and that's how I'm able to find different Things from different times But yeah the um if you're a digital hoarder, you're going to need a bigger storage device And any storage device you have should never be relied upon you should always also In addition to having a backup plan for that storage device not talked about that with synologies they actually From someone who's like if you're less technical synology is one of the easiest solutions to get started with deploy manage backup and Kind of set it and forget it. There's there's a lot less involved in synology than there is in some of the other ones It's not that I don't it's not that i'm trying to steer people away It's deciding whether or not you want to get into a more complicated solution I love true to ask that's what we run for all of our stuff and we do a lot of enterprise true to ask and we love it But if for someone's going hey, i'm a casual home user that wants to backup my phone And I just want to set up an app. Synology's really got a good done a great job in that market for that So definitely works well and this test is just keeps on going So close to being done Maybe I don't know it could be it could just sit here and run forever Oh look Now we can see the results and end the vlog It's now we're curious. Yes upload all the things and the winner is local mvme Beat it by a bit here 186,000 versus 133 This is not quite double that that mvme is really fast For this particular workload for these particular tests. So in this scenario, I would say mvme is faster But there's 12 drives. There's redundancy. There's More available storage on the Synology storage than there is in local mvme. So it's not exactly a clear winner because we go to our storage pools here the local Give it a good name. We have 400 gigs there, but if we go over to storage and we look at the Here we've got this much, but that's only because I've only allocated that much I can actually go here There's plenty more available. This storage pool actually has 4.2 terabytes, but when you go to the iSCSI manager We've only allocated this much on here and we could allocate Much more if we wanted to and uh Yes, leave Uh, if we allocated more then well, that would change that dynamic again. So we'd have more storage to work with Hopefully that makes sense over there I see line is finally bit the ball and so should I ask for editing? Yes. We talked about that earlier. Yes, he did Yeah, the cloud backup and Synology their cloud backup services reasonably priced and turnkey Uh, I actually I did a review of it. I'm we started deploying this for some of our customers that just like the Synology system It's been something we've done tests with we've done restores with I think they've done a good job They've made the pricing easy to understand and reasonable I do have a comparison to true nas, but if you're looking for an easy simple turnkey system Synology is going to be more turnkey than true nas. That's I mean, that's just what it is I can't help it. This is funny to me I don't know why I'm childish You need to really replace your q6600. Yeah Get some of them second gen fourth gen del opti plex. Yeah, there's a there's a lot of different systems you can build out there um It's it's always a debate because there's actually a lot of good surplus hardware you can find on places like ebay as well, so that's Definitely, um, lots to think about there And if you notice too, I got the spooky background set up for the halloween for those of you that didn't hear me say that at the very beginning Close all these windows they have open Actually, we're going to shut things down. So reopen So let's go here here Migrate this back over to local storage because i'm going to delete I'm done with this analogy for now Well, I have some more testing to do but that testing doesn't require this. So we just migrate this back over In two minutes, it'll be on i'll delete the storage pool and that's how we conclude benchmarking We just sit here and run these all day. It's part of my job Um, and that's how we make the videos on youtube that we make here Tom runs a bunch of tests or my staff runs some of the tests We make sure all the tests are verified We actually ran into a problem because the tests weren't what I expect them to be And then I realized it was just a plug got moved over to 10 gig There's one of the things I do is when we start looking at the tests, we kind of go back and verify the setup Well, I should have verified it beforehand, but I ended up verifying it afterwards Because one of the important aspects of this if we go to the host Is making sure It's all connected at 25 gigs And someone had plugged in and moved to cable and it was plugged into the 10 gig port And uh, yeah, that was bad. It caused it caused some skewed results ahead of me to rerun all those results again, which took a lot of time It's one of those, you know Measure once Measure once before you end up cutting multiple times. It's the same thing apply. Check all the connections Make sure they're all connected at what you think they're at. Don't just assume It's not like it's hard. I mean I always I do is go here and click click And look and go hey, what is it connected at? Oh, there we go. Look at that There we go. That's what you wanted to see 25 gigs awesome. That's that's the important part Uh, do you have any networking basic videos you've done? Any of them i've done a networking? I don't know how basic they are because that's a vague term Uh, I have not done anything like what a subnet is level networking, but there's so many other people that did I kind of skipped all that so Like I didn't start my network videos at the most basic level Because if you look there's already some popular Graphics done videos that are out there to get people started with a lot of the basics So I don't have any ones that are that way Does truenas have a way to do lvm snapshots? uh You cannot do lvm snapshots inside of truenas. It uses zfs snapshots. Yes, you can actually One thing you can do inside of truenas Is You can choose multiple targets when you do zfs replication So your replication will replicate your snapshots and you can have multiple targets So yes There's a way to do it. It's just different because you're doing it all in truenas So hopefully that either confuses or clarifies things I'm not sure which one So you can have multiples and you can even Have them going to like this goes from this server to this server But you can also change that up and have it going to More servers and when you do a replication inside of here, I have a video on replication by the way When you do replications it also replicates to All the snapshots all the features the data sets all the details is within the data set you're replicating end up on the other system So yes, you can completely have all those snapshots and it's kind of clever the way it works, but yes So it is very it's different Concept. Yes snapshot Like lvm, but very different the way it works in zfs overall Uh, the synology has the synology 25 gig card in it It I haven't any problems with it. Uh, I should make a list of all the cards. I bought a variety of cards to test I don't remember which ones they are right now I I can't remember. I have to look it up and I've completely forgotten there I all the ones I tested worked the same so I didn't have any problems with them Yeah, and like you said here if you want something really simple that just works backups and everything synology That's why we recommend so many of them Matter of fact, uh, among the things we're working on right now. I tweeted about this the other day Actually, where can I find that at? Um In my twitter probably Where did it go? Bring that over Um, we are working on a synology ha setup as well so This is uh Another project. We just wanted to play with some more synologies do some fun stuff with that show you how ha works So these are two ds 36 17s fully loaded with drives and uh, yeah, it's just Fun stuff we're playing with here at the office Because you don't want to see how it works. So the staff set this up configured it I'll walk through the configuration on there. Synology actually is kind of nice to give you the Their own little graphics for how it all works to show you what's working. What's not so Definitely pretty cool So even more videos coming on that. All right Is there any other final questions before I wander off and go play outside because it's so nice out Because it's beautiful day out Based on your experience with software and mechanical engineering and department of robotics is uh editing lead I What? I have no idea what you're talking about Do this right here Bash the like button That's what you do. So I'm going to go to second writing if you guys can click the like button before we leave Let's see Don't virtualize true nas maybe I need to make a video why you shouldn't virtualize true nas That's like a good video topic because it comes up all the time I have a video about how to virtualize it and in that video I say why it's a bad idea But maybe I should just make a video called why you shouldn't virtualize true nas Here's all the here's all the people that lost their data When they virtualized it because of problems and things like that So I don't recommend it people try it if it's a lab. Sure. It's not something I'd run in production So hopefully that makes sense. All right. Thanks everyone for coming. Thanks everyone that hit the like button Thanks for all of you that will hit the like button in any of your future. I'm going to take off and go outside Take this little pumpkin home because it's really cool And uh, that's about it. All right. See you all later In the forums and check out my analogy video I just posted if you want more information on some of the benchmarks now You see how I make the benchmarks and that video I talk about the results of them. All right. Thanks