WEBVTT 00:00:00.399 --> 00:00:08.120 way back before the Big Move I built a 00:00:04.920 --> 00:00:11.519 100 plus terabyte storage server to 00:00:08.120 --> 00:00:13.759 replace the awful store data on 00:00:11.519 --> 00:00:18.240 disconnected drives on a shelf in the bathroom system that we were rocking 00:00:16.119 --> 00:00:22.640 before thanks Seagate for the awesome drives and thanks 45 drives.com for the 00:00:20.640 --> 00:00:28.400 Rock and personalized storinator server but some of you may have noticed that I 00:00:25.640 --> 00:00:33.399 never followed up on the performance testing that I promised to do on that 00:00:30.359 --> 00:00:36.840 machine I was supposed to be showing off 00:00:33.399 --> 00:00:40.399 1 Gigabyte per second transfers with the 00:00:36.840 --> 00:00:44.290 10 GB network setup what gives well 00:00:40.399 --> 00:00:49.289 today we finally get the whole 00:00:51.840 --> 00:00:58.640 story the master case 5 by Cooler Master gives you the freedom to truly make your 00:00:56.399 --> 00:01:01.640 midtower PC case your own with a variety of modular parts and access 00:01:00.519 --> 00:01:07.240 check out the link in the video description to learn more so the short 00:01:04.040 --> 00:01:09.439 version is this in spite of 45 drives 00:01:07.240 --> 00:01:16.200 telling me that they had customers with similar configs saturating a 10 GB link 00:01:12.520 --> 00:01:19.080 or more I couldn't even get half of that 00:01:16.200 --> 00:01:22.880 and it made no sense really I did a lot of tinkering with this box before 00:01:20.880 --> 00:01:27.159 eventually deploying it different network cards different Drive 00:01:24.759 --> 00:01:33.680 configurations and finally got to the point where it was whether freas 00:01:29.960 --> 00:01:35.479 Hardware or pbac I had to roll it out 00:01:33.680 --> 00:01:39.360 because we needed to put our data somewhere and I was just going to have 00:01:37.159 --> 00:01:46.880 to live with the results that I got I mean I know I know poor lonus only has 00:01:42.840 --> 00:01:50.439 300 to 350 megabyte per second speeds to 00:01:46.880 --> 00:01:53.920 his over 100 terabytes of safe storage 00:01:50.439 --> 00:01:56.320 boohoo but this disrupted my plans for 00:01:53.920 --> 00:02:01.240 our storage infrastructure in a bigger way than you might think in addition to 00:01:58.880 --> 00:02:06.240 archiving old stuff to the server my intention was to have our daily use NZ 00:02:04.280 --> 00:02:11.560 the SSD one that you probably remember from this video doing nightly syncs or 00:02:09.440 --> 00:02:16.040 even hourly checkpoints if we could get away with it so we'd have two full 00:02:13.720 --> 00:02:22.400 copies of all of our mission critical data so I wanted the magnetic NZ to be 00:02:19.319 --> 00:02:24.319 fast enough to handle that and any 00:02:22.400 --> 00:02:31.560 random data that our editors needed to read from it from old projects which we 00:02:27.519 --> 00:02:34.440 were not able to do so while I've had 00:02:31.560 --> 00:02:40.080 four months to diagnose this and Ponder what could be wrong because it's had up 00:02:36.840 --> 00:02:42.519 to 60 terabytes of important data on it 00:02:40.080 --> 00:02:52.560 with nowhere else to offload that I've had no choice but to just Lim along at 00:02:46.360 --> 00:02:52.560 300 350 megabytes per second until 00:02:52.879 --> 00:02:59.280 today Seagate sent us 35 of their new 8 00:02:57.040 --> 00:03:04.799 terabyte Enterprise capacity drives and no these are not the shingled platter 00:03:02.080 --> 00:03:09.120 archival ones these are rocking ass capable of well in excessive 200 00:03:06.879 --> 00:03:12.560 megabytes per second transfer speeds rated at 2 million hours meantime 00:03:11.159 --> 00:03:19.879 between failure and with a 5year warranty to back it up proper Enterprise 00:03:16.879 --> 00:03:21.440 capacity drives so I immediately tore 00:03:19.879 --> 00:03:25.879 them out of their packaging and began building pyramids that no just kidding 00:03:24.360 --> 00:03:29.959 well actually okay I did build pyramids but but what I actually built with them 00:03:27.519 --> 00:03:36.200 after the pyramids was two additional servers each to hold a copy of our 60 00:03:33.200 --> 00:03:38.200 terabytes of data while I worked on the 00:03:36.200 --> 00:03:42.000 Vault so one of those machines is actually eventually going to be a Nas 00:03:39.720 --> 00:03:46.680 unit at my house and the other one is going to be an off-site backup server 00:03:44.120 --> 00:03:51.720 for this puppy but each of those will get their own videos later so with the 00:03:49.159 --> 00:03:57.079 data safely stored on a hardware raid 6 and on a software butter FS raid 5 each 00:03:54.920 --> 00:04:03.319 of those transfers took over a day by the way I wiped the frez and be began 00:04:00.120 --> 00:04:05.840 trying things so first I tried six Drive 00:04:03.319 --> 00:04:13.480 vevs since that's a more optimal number for ZFS 2 nope still shoddy transfer 00:04:09.239 --> 00:04:16.919 speeds next I tried 10 Drive vevs no 00:04:13.480 --> 00:04:21.239 difference again finally in desperation 00:04:16.919 --> 00:04:24.120 I tried a 27 Drive raid zero an 00:04:21.239 --> 00:04:30.199 experimental class configuration that no one should trust to hold any data no 00:04:26.720 --> 00:04:32.840 matter how amazing the drives are and 00:04:30.199 --> 00:04:38.240 same thing which after talking to the folks at 45 drives about my findings 00:04:35.160 --> 00:04:40.919 revealed that the issue is probably a 00:04:38.240 --> 00:04:45.160 software one because they've seen NFS shares just fly in a similar 00:04:43.440 --> 00:04:50.440 configuration to mine which doesn't do me any good because this is a Windows 00:04:47.240 --> 00:04:52.479 environment and we need SMB shares and 00:04:50.440 --> 00:04:55.639 so I had to keep investigating because if I'm going to be running around saying 00:04:53.919 --> 00:04:59.680 this Nas unit in these drives are capable of over a Gigabyte per second of 00:04:57.520 --> 00:05:02.680 transfer speed we use them here at l Media Group I mean I'm basically 00:05:01.000 --> 00:05:09.360 endorsing the things it's not good enough to me for 45 drives to see it in 00:05:06.320 --> 00:05:11.720 their lab I need to see it so I've been 00:05:09.360 --> 00:05:15.199 chatting a lot with the unraid guys ever since they helped us do the two Gamers 00:05:13.479 --> 00:05:19.280 one CPU project which you should definitely check out if you haven't 00:05:16.840 --> 00:05:25.720 already and they offered to spend some time configuring an experimental raid 00:05:22.440 --> 00:05:28.759 five butter FS array in un raid and 00:05:25.720 --> 00:05:32.440 tuning both the network settings as well 00:05:28.759 --> 00:05:34.960 as the SMB share settings so our initial 00:05:32.440 --> 00:05:39.479 test on a vanilla unraid server was frankly pretty ho hum actually fairly 00:05:37.199 --> 00:05:44.039 similar there's that poor SMB optimization outside of Windows 00:05:41.039 --> 00:05:46.360 platforms rearing its ugly head again 00:05:44.039 --> 00:05:51.639 some 4 kilobyte packet and jumbo frame tuning to the network card tuning of 00:05:48.440 --> 00:05:53.759 unraid networking configuration and boom 00:05:51.639 --> 00:06:00.080 that my friends is the cleanest 10 gigabit transfer that I've actually ever 00:05:57.039 --> 00:06:02.440 seen now not a lot of lime Tech 00:06:00.080 --> 00:06:05.880 customers are running 10 gig e but from their perspective I guess it's just 00:06:03.840 --> 00:06:11.360 valuable R&D for down the road when that gear becomes more common but I mean even 00:06:08.599 --> 00:06:15.960 then this is not the kind of config that most people will encounter even on 00:06:13.360 --> 00:06:20.319 unrated I actually don't intend to continue to run it like this uh butter 00:06:18.120 --> 00:06:26.000 FS raid five and raid six are both in the experimental stage but the good news 00:06:23.880 --> 00:06:30.080 here is that what I realized after running the slow freas configuration for 00:06:28.759 --> 00:06:36.400 so long was that generally speaking I don't need 00:06:33.160 --> 00:06:38.360 more than the 200 to 220 megabyte pers 00:06:36.400 --> 00:06:44.520 second transfer speeds that my individual drives are capable of in a 00:06:41.400 --> 00:06:46.560 normal unraid array and that the only 00:06:44.520 --> 00:06:51.560 thing that needs to be lightning fast performance-wise is the new footage and 00:06:49.360 --> 00:06:57.440 projects that we offload to it relatively little of which is created on 00:06:53.840 --> 00:06:59.080 a daily basis so we devised a new plan 00:06:57.440 --> 00:07:05.080 and to help us realize the new plan Kingston stepped up and offered to send 00:07:02.280 --> 00:07:13.160 us eight of their e50 Enterprise grade 480 gig ssds with power loss protection 00:07:09.639 --> 00:07:15.960 these drives will act as a 2 tbte RAID 00:07:13.160 --> 00:07:22.199 10 right cache that will be capable of the full 10 gbit transfer rate for fast 00:07:19.479 --> 00:07:27.720 updates throughout the day and that then flushes nightly to the hard drives when 00:07:25.120 --> 00:07:33.840 no one is using them all of this can be completely transparent to the user so 00:07:30.160 --> 00:07:35.639 the only time we'll ever see sub 1 GB 00:07:33.840 --> 00:07:41.400 per second transfers is when we're accessing cold data or when doing a 00:07:38.400 --> 00:07:44.080 massive dump of over 2 terabytes at a 00:07:41.400 --> 00:07:48.520 time another cool side note is that this might turn out to be a better way to 00:07:45.960 --> 00:07:52.919 leverage the extra horsepower that this op server is leaving on the table anyway 00:07:51.360 --> 00:07:58.319 because she never touches more than about 20% CPU usage so I could take a 00:07:56.240 --> 00:08:04.440 couple of cores and turn them into a network rendering box or game server or 00:08:01.360 --> 00:08:06.400 something else and on the subject then 00:08:04.440 --> 00:08:12.440 of our server being op I guess that brings us to the conclusion it turns out 00:08:08.720 --> 00:08:15.080 that the hardware is but SMB shares on 00:08:12.440 --> 00:08:19.520 non Windows platforms take some tuning and optimization that if you're willing 00:08:17.599 --> 00:08:23.120 to endure the dense documentation and condescending attitude of the freenas 00:08:21.199 --> 00:08:26.800 community you could probably achieve there but instead I ended up working 00:08:25.199 --> 00:08:32.080 directly with lime Tech to have baked into an upcoming release of unraid 6 and 00:08:28.759 --> 00:08:34.440 I'm super happy with the new 00:08:32.080 --> 00:08:38.440 config if you're building a mobile app and searching for a simple payment 00:08:36.479 --> 00:08:44.039 solution you might want to check out brain tree with the brain tree v.0 SDK 00:08:41.760 --> 00:08:49.160 which is just one small snippet of code you can be all set up and ready in less 00:08:46.760 --> 00:08:51.920 than 10 minutes to take online payments and if you're having any trouble they 00:08:50.240 --> 00:08:55.399 have support staff ready to walk you through the process over the phone if 00:08:53.640 --> 00:09:00.800 you need them their code supports Android iOS and JavaScript clients and 00:08:58.440 --> 00:09:05.839 they have sdks in in seven programming languages and it makes it easy to offer 00:09:03.279 --> 00:09:12.920 multiple Mobile payment types including PayPal Apple pay Bitcoin venmo cards and 00:09:10.440 --> 00:09:15.760 more all with a simple integration they've got quick knowledgeable 00:09:14.519 --> 00:09:19.320 developer support if you have any questions and to learn more all you got 00:09:17.920 --> 00:09:25.200 to do is go over to Braintree payments.com Linus and if you use that link you can 00:09:23.600 --> 00:09:30.240 also get your first $50,000 in transactions with no fees 00:09:28.640 --> 00:09:34.519 whatsoever so check it out today at the link in the video description so thanks 00:09:32.399 --> 00:09:38.560 for watching guys if this video sucked you know what to do but if it was 00:09:36.040 --> 00:09:43.839 awesome get subscribed hit the like button like button hit the like button 00:09:42.200 --> 00:09:48.040 or even consider supporting us directly by using our affiliate code to shop at 00:09:46.320 --> 00:09:51.480 Amazon instructions for which we're up here buying a cool shirt like this one 00:09:49.880 --> 00:09:55.680 or with a direct monthly contribution through our form it gets you a little contributor tag now that you're done 00:09:54.640 --> 00:10:00.920 doing all that stuff you're probably wondering what to watch next so click that little button in the top right 00:09:58.720 --> 00:10:06.760 corner to check out the ultimate showdown between AMD's 8 core CPU and 00:10:04.200 --> 00:10:12.120 Intel's 8 core server anyway it's complicated you guys should check it out 00:10:08.160 --> 00:10:12.120 the video is there