1 00:00:00,000 --> 00:00:00,900 In this lesson, 2 00:00:00,900 --> 00:00:02,940 we're going to cover high availability. 3 00:00:02,940 --> 00:00:04,920 These days, our organizations expect 4 00:00:04,920 --> 00:00:07,830 to have our services up and running pretty much non-stop. 5 00:00:07,830 --> 00:00:09,150 And the only way to achieve that 6 00:00:09,150 --> 00:00:10,620 is by creating an architecture 7 00:00:10,620 --> 00:00:13,080 that will support high levels of availability. 8 00:00:13,080 --> 00:00:15,690 Now, high availability refers to the ability of a service 9 00:00:15,690 --> 00:00:18,360 to be continuously available by minimizing the downtime 10 00:00:18,360 --> 00:00:20,070 to the lowest amount possible. 11 00:00:20,070 --> 00:00:21,570 To achieve high availability, 12 00:00:21,570 --> 00:00:23,640 our systems should be designed using load balancing 13 00:00:23,640 --> 00:00:26,280 or clustering, have redundancy built into the system, 14 00:00:26,280 --> 00:00:28,770 and if we're operating in a cloud-based environment, 15 00:00:28,770 --> 00:00:31,290 we should also consider using a multi-cloud environment 16 00:00:31,290 --> 00:00:32,700 to achieve an operational environment 17 00:00:32,700 --> 00:00:34,590 that can withstand multiple points of failure 18 00:00:34,590 --> 00:00:36,900 before our service becomes unavailable. 19 00:00:36,900 --> 00:00:38,670 Now, whenever we talk about availability, 20 00:00:38,670 --> 00:00:40,680 it's important to remember that we measure our levels 21 00:00:40,680 --> 00:00:44,010 of availability in terms of something known as uptime. 22 00:00:44,010 --> 00:00:45,810 Uptime is the number of minutes or hours 23 00:00:45,810 --> 00:00:48,570 that your system remains online over a given period, 24 00:00:48,570 --> 00:00:50,640 and this uptime is usually going to be expressed 25 00:00:50,640 --> 00:00:51,960 as a percentage. 26 00:00:51,960 --> 00:00:54,060 The gold standard in availability is set 27 00:00:54,060 --> 00:00:55,170 by achieving something known 28 00:00:55,170 --> 00:00:57,390 as the five nines of availability. 29 00:00:57,390 --> 00:00:59,310 The five nines of availability refers 30 00:00:59,310 --> 00:01:02,580 to achieving a 99.999% uptime, 31 00:01:02,580 --> 00:01:04,830 which means that your services can only have a maximum 32 00:01:04,830 --> 00:01:07,110 of about five minutes of downtime per year, 33 00:01:07,110 --> 00:01:08,790 which as you can probably imagine, 34 00:01:08,790 --> 00:01:10,800 isn't a whole lot of downtime. 35 00:01:10,800 --> 00:01:12,390 Now, in some cloud-based networks, 36 00:01:12,390 --> 00:01:14,250 organizations have gone even further 37 00:01:14,250 --> 00:01:16,860 by setting their availability target at six nines 38 00:01:16,860 --> 00:01:20,760 of availability, or 99.9999%, 39 00:01:20,760 --> 00:01:23,400 which equates to only 31 seconds of downtime 40 00:01:23,400 --> 00:01:25,050 each and every year. 41 00:01:25,050 --> 00:01:26,160 Now, as you can imagine, 42 00:01:26,160 --> 00:01:27,810 most organizations are going to need 43 00:01:27,810 --> 00:01:30,720 to take their systems offline for more than 31 seconds 44 00:01:30,720 --> 00:01:33,780 per year and probably more than five minutes each year too 45 00:01:33,780 --> 00:01:36,240 in order to install security patches on their servers 46 00:01:36,240 --> 00:01:37,890 or to replace a failed hard drive 47 00:01:37,890 --> 00:01:40,320 or to install a new router or switch into the network 48 00:01:40,320 --> 00:01:42,060 when the old one has failed. 49 00:01:42,060 --> 00:01:44,790 So how can we maintain a high level of availability 50 00:01:44,790 --> 00:01:47,070 while still being able to take down some of our systems 51 00:01:47,070 --> 00:01:49,260 to perform our maintenance and repairs? 52 00:01:49,260 --> 00:01:51,000 Well, the first way we can do this in order 53 00:01:51,000 --> 00:01:52,860 to achieve higher levels of availability 54 00:01:52,860 --> 00:01:54,687 is to prevent our systems from becoming overloaded 55 00:01:54,687 --> 00:01:56,820 by using load balancing. 56 00:01:56,820 --> 00:01:59,190 Load balancing is the process of distributing workloads 57 00:01:59,190 --> 00:02:01,140 across multiple computing resources 58 00:02:01,140 --> 00:02:04,230 to optimize the resource use, maximize their throughput, 59 00:02:04,230 --> 00:02:05,670 minimize their response time, 60 00:02:05,670 --> 00:02:08,520 and prevent the overloading of any single resource. 61 00:02:08,520 --> 00:02:11,550 Load balancers achieve this by using complex algorithms 62 00:02:11,550 --> 00:02:13,470 to distribute incoming requests to servers 63 00:02:13,470 --> 00:02:14,850 that are capable of fulfilling them 64 00:02:14,850 --> 00:02:16,770 efficiently and effectively. 65 00:02:16,770 --> 00:02:18,990 For example, if I'm running a small website 66 00:02:18,990 --> 00:02:21,270 to host my blog with only a few dozen readers, 67 00:02:21,270 --> 00:02:23,790 I can probably get by with just having a single server 68 00:02:23,790 --> 00:02:25,980 that can handle that amount of simultaneous requests 69 00:02:25,980 --> 00:02:27,000 that I'm expecting to receive 70 00:02:27,000 --> 00:02:29,160 without causing my server to crash. 71 00:02:29,160 --> 00:02:31,530 But if my blog starts to gain popularity 72 00:02:31,530 --> 00:02:33,240 and now I have a few thousands people who are trying 73 00:02:33,240 --> 00:02:35,790 to read it, a single server is going to be insufficient 74 00:02:35,790 --> 00:02:37,680 to handle all of that additional load. 75 00:02:37,680 --> 00:02:40,470 So instead, I can set up two or three servers 76 00:02:40,470 --> 00:02:42,720 that will all take turns responding to my visitors, 77 00:02:42,720 --> 00:02:45,840 to be able to ensure that no single server gets too overloaded. 78 00:02:45,840 --> 00:02:47,970 This is what load balancing allows us to do, 79 00:02:47,970 --> 00:02:50,580 because all the requests go to the load balancer first, 80 00:02:50,580 --> 00:02:53,400 and then the load balancer will redirect that user's request 81 00:02:53,400 --> 00:02:55,620 to one of my three servers in order to respond 82 00:02:55,620 --> 00:02:57,780 to the request in a more timely manner. 83 00:02:57,780 --> 00:02:59,790 Now, second, we have clustering. 84 00:02:59,790 --> 00:03:01,560 Clustering can be set up to help your systems 85 00:03:01,560 --> 00:03:04,020 to handle more traffic at the same time. 86 00:03:04,020 --> 00:03:06,240 Clustering refers to the use of multiple computers, 87 00:03:06,240 --> 00:03:09,030 multiple storage devices, or redundant network connections 88 00:03:09,030 --> 00:03:11,250 that all work together as a single system 89 00:03:11,250 --> 00:03:13,020 to provide higher levels of availability, 90 00:03:13,020 --> 00:03:16,560 reliability, and scalability for your enterprise systems. 91 00:03:16,560 --> 00:03:19,140 Clustering is more about keeping an application available 92 00:03:19,140 --> 00:03:21,000 even in the event of a hardware failure 93 00:03:21,000 --> 00:03:21,833 in order to ensure 94 00:03:21,833 --> 00:03:24,000 that there is no single points of failure. 95 00:03:24,000 --> 00:03:26,550 While load balancing can help handle all the excess traffic 96 00:03:26,550 --> 00:03:29,460 by distributing it, clustering will provide redundancy 97 00:03:29,460 --> 00:03:31,110 in the event of a system failure 98 00:03:31,110 --> 00:03:32,910 to ensure that the continuity of service is going 99 00:03:32,910 --> 00:03:34,200 to be maintained. 100 00:03:34,200 --> 00:03:36,780 Load balancing and clustering can also be combined together 101 00:03:36,780 --> 00:03:38,160 in the same architecture 102 00:03:38,160 --> 00:03:39,750 to provide an even more robust system 103 00:03:39,750 --> 00:03:42,150 for maintaining higher levels of availability. 104 00:03:42,150 --> 00:03:43,980 Under this type of combined design, 105 00:03:43,980 --> 00:03:46,590 load balancing will manage and optimize your resources 106 00:03:46,590 --> 00:03:47,880 under normal conditions, 107 00:03:47,880 --> 00:03:49,470 but then clustering can take over 108 00:03:49,470 --> 00:03:51,690 to ensure the service remains available in the event 109 00:03:51,690 --> 00:03:54,390 of a component failure or a system failure. 110 00:03:54,390 --> 00:03:56,910 Third, our high availability systems should be built 111 00:03:56,910 --> 00:03:58,530 with redundancy in mind. 112 00:03:58,530 --> 00:04:01,050 Now, redundancy is the duplication of critical components 113 00:04:01,050 --> 00:04:02,460 or functions of a system 114 00:04:02,460 --> 00:04:04,560 with the intention of increasing the reliability 115 00:04:04,560 --> 00:04:05,970 of a given system. 116 00:04:05,970 --> 00:04:07,920 To create a highly available infrastructure, 117 00:04:07,920 --> 00:04:09,810 redundancy has to be built into your designs 118 00:04:09,810 --> 00:04:12,270 by installing or adding multiple power supplies, 119 00:04:12,270 --> 00:04:14,850 network connections, servers, software services, 120 00:04:14,850 --> 00:04:17,760 or service providers to your architecture's design. 121 00:04:17,760 --> 00:04:19,500 By using redundant power supplies, 122 00:04:19,500 --> 00:04:21,600 you can ensure that the failure of one power source 123 00:04:21,600 --> 00:04:24,480 is not going to affect the continuity of your other services. 124 00:04:24,480 --> 00:04:26,490 Now, this type of power supply redundancy 125 00:04:26,490 --> 00:04:28,590 can be achieved by installing two power supplies 126 00:04:28,590 --> 00:04:30,270 inside of a server directly, 127 00:04:30,270 --> 00:04:32,550 or it can be achieved by ensuring your facility 128 00:04:32,550 --> 00:04:34,110 has multiple power sources 129 00:04:34,110 --> 00:04:36,870 by using an uninterrupted power supply, or UPS, 130 00:04:36,870 --> 00:04:39,210 a backup generator, or maintaining connections 131 00:04:39,210 --> 00:04:41,520 to two or more power grids simultaneously 132 00:04:41,520 --> 00:04:43,140 to ensure your systems always have access 133 00:04:43,140 --> 00:04:45,090 to the power they need to operate. 134 00:04:45,090 --> 00:04:47,010 In addition to redundant power supplies, 135 00:04:47,010 --> 00:04:47,843 it's also important 136 00:04:47,843 --> 00:04:50,280 that you maintain multiple network connections or pathways 137 00:04:50,280 --> 00:04:51,960 by using multiple cabled connections 138 00:04:51,960 --> 00:04:54,630 or a cable connection and a wireless connection 139 00:04:54,630 --> 00:04:56,550 in order to prevent any single points of failure 140 00:04:56,550 --> 00:04:59,670 from disrupting your service to your network's connectivity. 141 00:04:59,670 --> 00:05:01,920 Now, another form of redundancy comes in the form 142 00:05:01,920 --> 00:05:03,360 of servers and services, 143 00:05:03,360 --> 00:05:05,370 which can be configured to operate in a load balanced 144 00:05:05,370 --> 00:05:07,766 or clustered architecture to help prevent downtime 145 00:05:07,766 --> 00:05:10,050 by ensuring that a backup will always be available 146 00:05:10,050 --> 00:05:11,520 in case one of your servers suffers 147 00:05:11,520 --> 00:05:13,230 from a catastrophic failure. 148 00:05:13,230 --> 00:05:15,690 Similarly, deploying redundant software services 149 00:05:15,690 --> 00:05:17,430 or components can be used to ensure 150 00:05:17,430 --> 00:05:18,990 that if one instance fails, 151 00:05:18,990 --> 00:05:20,700 then the remaining workload can be shifted 152 00:05:20,700 --> 00:05:22,680 to another instance seamlessly. 153 00:05:22,680 --> 00:05:24,330 Now, additionally, if you're reliant 154 00:05:24,330 --> 00:05:25,590 on external service providers 155 00:05:25,590 --> 00:05:27,570 to provide a critically important service to you, 156 00:05:27,570 --> 00:05:29,490 like your internet service provider does, 157 00:05:29,490 --> 00:05:31,380 then you should probably protect your organization 158 00:05:31,380 --> 00:05:34,350 from an outage by using two or more service providers 159 00:05:34,350 --> 00:05:36,810 so that you always have a readily available backup. 160 00:05:36,810 --> 00:05:38,580 This kind of provider diversity means 161 00:05:38,580 --> 00:05:40,920 that if one provider experiences an internal issue, 162 00:05:40,920 --> 00:05:43,380 the organization's operations will be able to continue 163 00:05:43,380 --> 00:05:45,570 by using another provider's services. 164 00:05:45,570 --> 00:05:47,640 For example, at DionTraining.com, 165 00:05:47,640 --> 00:05:50,010 we use a credit card service provider known as Stripe 166 00:05:50,010 --> 00:05:51,510 to collect payments from our students 167 00:05:51,510 --> 00:05:53,940 when they buy their exam vouchers on our website. 168 00:05:53,940 --> 00:05:56,520 But we also have a secondary provider already configured 169 00:05:56,520 --> 00:05:58,890 and ready for us to enable if Stripe has some kind 170 00:05:58,890 --> 00:06:00,540 of a catastrophic outage. 171 00:06:00,540 --> 00:06:02,490 Normally, we prefer to rely on Stripe 172 00:06:02,490 --> 00:06:04,260 because they provide a great level of service 173 00:06:04,260 --> 00:06:06,510 at a lower price than our secondary provider. 174 00:06:06,510 --> 00:06:08,820 But if Stripe goes offline for some reason, 175 00:06:08,820 --> 00:06:09,750 we're already configured 176 00:06:09,750 --> 00:06:12,240 that our backup credit card processor can take over 177 00:06:12,240 --> 00:06:14,730 until our Stripe-based service is once again ready 178 00:06:14,730 --> 00:06:17,340 to start conducting normal operations again. 179 00:06:17,340 --> 00:06:19,650 Another example might be having two domain controllers 180 00:06:19,650 --> 00:06:21,570 in your enterprise network environment. 181 00:06:21,570 --> 00:06:23,040 Under this kind of a design, 182 00:06:23,040 --> 00:06:24,390 you can set up your domain controllers 183 00:06:24,390 --> 00:06:26,550 as a primary and secondary controller. 184 00:06:26,550 --> 00:06:29,040 And if you need to restart one to apply a security patch, 185 00:06:29,040 --> 00:06:31,350 the other controller can continue to maintain the services 186 00:06:31,350 --> 00:06:33,030 for all of your end users. 187 00:06:33,030 --> 00:06:34,950 Having redundant hardware for everything 188 00:06:34,950 --> 00:06:36,540 would become really expensive, 189 00:06:36,540 --> 00:06:38,220 because effectively, we are doubling the cost 190 00:06:38,220 --> 00:06:39,360 to build out our network. 191 00:06:39,360 --> 00:06:41,970 So it's important that each design decision is considered 192 00:06:41,970 --> 00:06:43,710 to decide if redundancy is required 193 00:06:43,710 --> 00:06:45,240 for a particular piece of hardware 194 00:06:45,240 --> 00:06:47,370 or if you could build the desired level of redundancy 195 00:06:47,370 --> 00:06:50,280 by using software or cloud-based services instead. 196 00:06:50,280 --> 00:06:52,230 Either way, though, by adding additional layers 197 00:06:52,230 --> 00:06:54,360 of redundancy to your overall architecture, 198 00:06:54,360 --> 00:06:55,650 you're going to be able to help minimize 199 00:06:55,650 --> 00:06:57,750 the potential points of failures in your systems 200 00:06:57,750 --> 00:07:00,390 and ensure that your system or services remain available 201 00:07:00,390 --> 00:07:02,580 despite any unforeseen issues. 202 00:07:02,580 --> 00:07:04,890 Fourth, we can utilize multi-cloud systems 203 00:07:04,890 --> 00:07:07,470 to distribute our data, applications, and services 204 00:07:07,470 --> 00:07:09,570 across several different cloud-based environments 205 00:07:09,570 --> 00:07:11,610 as an additional form of redundancy. 206 00:07:11,610 --> 00:07:13,740 By implementing a multi-cloud architecture, 207 00:07:13,740 --> 00:07:15,960 you can mitigate the risk of a single point of failure, 208 00:07:15,960 --> 00:07:17,430 because if one of your cloud providers 209 00:07:17,430 --> 00:07:18,600 experiences an outage, 210 00:07:18,600 --> 00:07:20,040 then your organization's workload 211 00:07:20,040 --> 00:07:22,650 can be rapidly transitioned to another cloud provider 212 00:07:22,650 --> 00:07:25,260 with minimal disruptions to your operations. 213 00:07:25,260 --> 00:07:26,730 Using a multi-cloud system 214 00:07:26,730 --> 00:07:28,710 will also provide you with some additional flexibility 215 00:07:28,710 --> 00:07:31,650 for scaling your operations and for optimizing your costs, 216 00:07:31,650 --> 00:07:33,030 because different service providers 217 00:07:33,030 --> 00:07:36,060 may offer different pricing structures for similar services. 218 00:07:36,060 --> 00:07:38,220 So you can opt to use the least expensive one 219 00:07:38,220 --> 00:07:40,200 that meets your needs as your primary provider 220 00:07:40,200 --> 00:07:42,720 for a given service while maintaining the other provider 221 00:07:42,720 --> 00:07:46,320 as your secondary or backup provider for that same service. 222 00:07:46,320 --> 00:07:49,110 A multi-cloud design also helps your organization avoid 223 00:07:49,110 --> 00:07:51,210 any kind of potential vendor lock-in issues, 224 00:07:51,210 --> 00:07:53,700 because it provides you with more options and leverage 225 00:07:53,700 --> 00:07:56,160 when it comes time to negotiate your service terms 226 00:07:56,160 --> 00:07:57,840 or if you need to migrate your services 227 00:07:57,840 --> 00:08:00,000 to another cloud provider due to an outage 228 00:08:00,000 --> 00:08:01,710 or other types of issues. 229 00:08:01,710 --> 00:08:04,350 However, while implementing a multi-cloud strategy, 230 00:08:04,350 --> 00:08:05,190 it's going to be important 231 00:08:05,190 --> 00:08:06,990 that you ensure proper data management, 232 00:08:06,990 --> 00:08:08,100 unified threat management, 233 00:08:08,100 --> 00:08:10,170 and consistent policy enforcement is being used 234 00:08:10,170 --> 00:08:12,150 across all of your cloud-based environments 235 00:08:12,150 --> 00:08:14,610 so that your organization can maintain a higher security 236 00:08:14,610 --> 00:08:16,200 and compliance posture. 237 00:08:16,200 --> 00:08:18,780 So remember, to achieve high availability, 238 00:08:18,780 --> 00:08:20,670 you're going to need to conduct some strategic planning 239 00:08:20,670 --> 00:08:22,950 in order to design a more robust system architecture 240 00:08:22,950 --> 00:08:24,360 for your organization. 241 00:08:24,360 --> 00:08:26,250 By understanding and implementing load balancing 242 00:08:26,250 --> 00:08:28,860 and clustering, instituting redundancy and multiple layers 243 00:08:28,860 --> 00:08:30,300 of the operational framework, 244 00:08:30,300 --> 00:08:32,309 and utilizing a multi-cloud approach, 245 00:08:32,309 --> 00:08:34,559 organizations can significantly reduce their risk 246 00:08:34,559 --> 00:08:37,890 of service disruptions and the associated downtime costs. 247 00:08:37,890 --> 00:08:40,080 This kind of proactive approach helps to safeguard 248 00:08:40,080 --> 00:08:43,049 your organization's operational continuity and reliability 249 00:08:43,049 --> 00:08:45,540 while also fortifying its reputation and credibility 250 00:08:45,540 --> 00:08:48,290 in an increasingly competitive operational environment.