 Tom here from Orange Systems and let's talk about these risings I have in the rack here. So this has been a project and ongoing process and tearing apart my old what used to be my studio. If you remember there's all the different rails and stuff that held the lights and turning it into the area where we have some of our lab and all of our demos that we set up and things for YouTube. And then we have these risin systems here running XCPNG. Now one of the things I want to talk about with these is there's not much to see inside of them but I'll open it up because people will want to know. We do keep them locked so it's you know looks nicer and have the USBs blocked off that aren't any use in there because I don't really use them much. These two systems I'll have a full parts description down below. They run XCPNG and then down on the bottom here and I just reviewed this. This is my 45 drives Q30. Now one of the questions people had asked is can you still slide this out and of course the answer is yes. And why are they asking that question? Well let's go around to the back and find out. So here is the cabling and this is almost done so I'm not going to spend a lot of time talking about this. Well that'll be a later video I do but let's get down here and talk about what is cable. Now the two risin servers here are cabled with the orange being our storage network and then we have the black here is 25 gig and that's the data series network. Now we do have one other network over here which you have our IPMI and then we have one more network that we keep physically separate where we have things a little more secure and I don't want them to just v-land off so they're not sharing any traffic and one gig is perfectly adequate for what runs on there. Now neither one of these are on rails these are just sitting here which will be a trigger moment for at least a few people leave the comments down below on that. But they were talking about the tightness of these. These are just SFPs so if I have to slide this out not often I would. We'll get the details of what hardware is in there and why there's nothing really hot swappable so I don't really need to slide them out to get to things but I can just unplug these couple things that are color coded easy to remember where they are. Now as I said the 45 drive server yes it slides out perfectly fine. There's enough cable slack to accommodate that so not really a big deal from wiring. Now I have the full parts list listed down below but I want to start with talking about this Melanox and the Q-Logic cards that I have in here because card compatibility comes up a lot with XCPNG which ones work. Well both of these work perfectly fine with a really minor difference that I don't know exactly where the problem is but at least I'll demonstrate it here. First let's talk about how it's connected. So the black cables and then the orange cables the orange cables represent the storage network and those are actually only connected at 10. You're probably thinking wouldn't it make more sense to have the storage at the full 25 and the reason I didn't is because well our 45 drive server wouldn't be able to saturate that. Also it doesn't have a 25 gig adapter in it at the moment and because it's just a spinning rust server I'll see the review down below it doesn't really it's not likely to actually fill up that entire pipe of 25 gig. So these are all connected at 10 gig then the other network interface is essentially for the data network has a lot of VLANs has quite a few things running on it and that one is connected at 25. So two ports on our usw pro aggregation switch are connected at 25 then there's a fiber connection that goes over to another usw pro aggregation switch is connected to some more servers that I have that also one of them has a 25 gig adapter and that's we're going to do a quick demo on this. So the systems here and this one right here where we're seeing the 18.1 gigs we can do it it's really consistent. This is the one with the Q logic card in there it really consistently gets just about 18.2 or 18.3 kind of ounces around in there and that's the best I can get out of it but if we use the milanox card it pretty consistently gets 20 21 kind of bounces back and forth between there now there's a few other services running on it so it's not like I'm testing this only and natively with this but it's kind of strange I don't get completely 25 but I've also not spent a lot of time testing it now I do know the switch itself is capable of 25 I've tested it with some other just set up a couple raw servers and pump data back and forth but in production with everything attached to it with all the VLANs everything configured these are the results I'm getting out of here maybe I'll spend some time digging in to see if there's some fine tuning but for the most part I'm happy with the 18.3 and 20.5 gigs a second that I respectively get out of these systems here and here is the ASRock IPMI dashboard the 79 days is accurate for how long we've had this thing running and doing testing with it it hasn't given us any trouble it's worked quite well with xcp and g I like that it's accumulative power on ours it's not necessarily uptime of you know since its reboot or since its last power cycle it's just a cumulative number of hours that this has been running I also like that right here we can jump to the access logs and see who's been logging in and it's apparently this guy Tom and it also keeps track of the KVM so let's just jump right over to the remote control KVM HTML5 viewer it loads this fast and I'm doing this over a VPN I'm here in my studio this is at my office and I mean we have a fast connection between here but it's pretty immediate when you want to get to the KVM you're not waiting around for it now I didn't really spend any time testing it does have the ability to browse and upload different cd images so you don't even have to plug a usb in to get this set up there's also the ability to map network resources to be able to boot this up off of different images not something I spent a lot of time testing maybe if I get some time I'll come back and revisit that to take a look at more in depth how that works but you do have all the sensor information since system information inventory power source it's pretty straightforward the only thing it can be a little bit confusing is how you may want to set up the settings in terms of the users it's kind of kind of a weird way of doing user management so if we go over here to the user management part you have to understand that it's all broken down into different channels and how you set it up not to be a deal once you stare at a minute but you can set up an administrator but you can't disable administrator but you can change it to a really crazy long password and then remove some of the permissions so it's not an normal administrator login so you're using a different login than just administrator on there I know that's a little bit of security through obscurity especially from a guy on YouTube who just shared the username that he is logging in with which you'll probably have to change this video now because well you know Tom admin seems like a pretty obvious guess anyways but overall it's been pretty pain free doing this I like the fact that you have the backup configuration preserve configuration restore so once you get all the settings set you can back them all up and in case you have to replace some other board you could just put all the settings back in now one more thing I haven't set up yet is the remote logging but I like that this is an option so we can actually have this board directly sending information to a logging server and with the logging server you know you can get a little bit more fine-grain detail and if there's something going wrong hopefully the logging server you can trigger on some alerts to catch that before there's a real problem and that's the last thing I want to talk about is because I set the drives up in this as mirrored drives of mirrored MVMEs and you're probably asking well how do you monitor mirrored MVMEs because XCP and G and Zanorka don't have the facility to do that well let's show you how I do that it's actually pretty simple I mean you're loading XCP and G you can set it up on a mirrored radar so this is a mirrored MVME you can take either one out it will still boot but of course what happens when it fails how do you know about it well instead of having the servers themselves notify me I actually have the server sending everything to a syslog server gray log I've been using gray log to ingest all the logs across all of my different systems and then you build triggers on it so right now this is happy and there's nothing wrong with the raid so when you do the MDADM-D the device ID you can see that everything's fine now maybe I'll do a separate video if there's enough comments down below asking how I did this but it's really quite simple we're going to go over here and look at gray log and this is the alert definition XCP and G raid failure event definition that sends a notice it really simple search query for disk failure that will get dumped into the XCP and G logs and because the logs are piped over to gray log gray log will let me know which server failed because it'll be part of the log definition so you just parse for that and it checks every minute so if one of these were to fail it would immediately have gray log grab that send me an email and you know bring me to action to go hey that's interesting I've have a failure on whichever server so tell me which server it came from and then I can make the choice of what I want to do with that piece of information which is probably start migrating things off that host figure out why the disk failed and replace the failing drive in it so far that hasn't happened but that's all you really have to do is look for some disfail your notices not any more complicated than that now that we've covered all the hardware let's talk about performance this performs really well without any hiccups I haven't had any issues using the rise and processors now the way you do this in Zen server is you're going to create a resource pool now even an individual host is technically in its own resource pool by itself you put more than one host in a resource pool they can use things like shared storage that allows you to store the VMs on a storage target such as my 45 drives running true nas and then the VM can start on either one of these hosts now that does not mean it's in a j mode a j does require three or more devices that way if one of them fails they can have a quorum and decide where to start so we just have two and when you want to move them I'll show you how fast that moves the 25 gig connection between them makes it really easy to take and running machines and just swap them over provide they're on the shared storage target local storage so I'll mention it does have the MVMEs as local storage as well you can use them for both booting and for the extra being local storage now they're just in a mirror but still their MVMEs so the performance is pretty good on those and as I noted with the alerting if there's a problem or one of them goes out not a big deal and you can live migrate storage so even though these are running virtual machines that are in here I can live migrate them if I think there's a problem to the other host or to the shared storage between the hosts without a problem now the rise in systems people ask a lot about this I've seen these comments going well Tom shouldn't you have went with a faster processor something you know substantially faster and the reality is and we go over here to the stats and we look at maybe well even the last 10 minutes here a little bit of testing has brought our CPU spiking to 29% but if we spread it out like over let's say a week there's a few spikes but very few spikes we're not even fully utilizing this system to its full potential this is one of the reasons we didn't go with like a whole epic system or something even more high-end now let's actually look at some of the VMs running and show you how fast they can be migrated between two different systems so currently this is my cali linux box which has eight cores assigned to it and 16 gigs of memory it's running on server rise in two let's say I want to move it over because it's using shared storage the trinity pool of zen shared storage server right here it's an nfs connection and you can see that it's got a browser pulled up so it's doing something not much just booted and let's just migrate it real quick over to the rise in one server now it'll figure out the migration network to use the 25 gig network by default so let's hit okay and actually we'll jump over to your task and I'm doing this on real time and none of this is sped up and we are let's see 11 10 9 8 it's actually going a little faster so I think this process will probably take well it looks like probably under 10 seconds I didn't really pull a timer up for that but now we go back over here and look it's still running perfectly fine it zoom it back out to normal and now it's on the other server so you can do these migrations just absolutely fast that's not a problem that seemed to be a concern few people have like will the virtualization work will it be able to do live migrations of bm's all that works perfectly fine with these particular rise in processors no problem there I mean if we go back over here and we have something like the pharaonex server which is a little bit smaller with only eight gigs ram I can do the same thing it'll move even faster now also in once again it's in a shared resource pool so we're going to go ahead and take it we can stop it and then if you're wondering you can start it on either server right here because it's in that shared pool so if we started it on rise in one go ahead and let it start up we'll run through this real fast actually we'll let it do it in real time you're giving an idea of just how fast these boot up so probably it's going to take about 10 seconds maybe 15 the latest sabon 2204 does do this it has a blank screen where you're scratching your head going is it doing something is it stuck it is definitely doing something you'll see it kind of pause right here and by time it goes back hey now it's got a screen and this is all still real time nothing sped up almost booted and come on there we go completely ready now it looked for cloud or next I didn't turn that off on this particular virtual machine but now that we spun it up on here let's just migrate it over to the other server real quick hit okay hit okay same thing and with only eight gigs to move between the virtual machines it moves even faster over there and I don't know was it probably about six or seven seconds to get that moved over so my overall with the performance on it I've been really happy really impressed with it and if I need to move something to one of these local storage I can just choose this and then choose the local storage that I would want to move it to and I can do that live as well I've moved things back and forth I haven't really run a hiccups at all matter of fact one of the things compared to the Dell R630 this was on is the single thread performance on these Ryzen is so much faster and a lot of the disk IO is dependent on single thread performance so we got quite a bit of performance boost with this system it's actually impressively fast with IO not just for shared storage but even with the local MVME the overall increased performance has been really nice for setting up all these VMs. I still have another server we're probably going to build because I think we might want one more if we build this other project idea I have we use this a lot for our testing of all the different things that sometimes I demo here or times when we have to build things for our clients the other thing we use it for is some DR testing because untested backups are just wishful thinking so we'll on behalf of our clients do certain DR tests have their files make sure we can rebuild their network without their network make sure we can rebuild aspects of it that's why this system seemed to be reasonably fast but they're not under high load all the time just when we have these little projects that come through or I want to lab something out or set up a demo for YouTube but leave your thoughts comments and questions down below or head up to our forums for a more in-depth discussion and you know let me know what you think of this build and what you would have done differently because I'm positive many people have already commented about what they would change and that's fun because I think it opens up for a lot of discussions I will never claim I am a absolute hardware expert on choices some of it was driven by what was on sale what we could get a deal on what we thought was reasonable and of course one really important aspect of this is what we found stable because obviously crashing uh really inhibits getting work done all right and thanks and thank you for making it all the way to the end of this video if you've enjoyed the content please give us a thumbs up if you would like to see more content from this channel hit the subscribe button and the bell icon if you like to hire a sure project head over to laurancesystems.com and click the hires button right at the top to help this channel out in other ways there's a join button here for youtube and a patreon page where your support is greatly appreciated for deals discounts and offers check out our affiliate links in the description of all of our videos including a link to our shirt store where we have a wide variety of shirts that we sell and designs come out well randomly so check back frequently and finally our forums forums.laurancesystems.com is where you can have a more in-depth discussion about this video and other tech topics covered on this channel. Thanks again for watching and look forward to hearing from you.