WEBVTT 1 00:00:00.060 --> 00:00:01.350 In this lesson, 2 00:00:01.350 --> 00:00:04.860 we will learn about Performance Considerations. 3 00:00:04.860 --> 00:00:08.070 Performance considerations balance the need 4 00:00:08.070 --> 00:00:09.690 for strong security 5 00:00:09.690 --> 00:00:14.040 with the efficiency and speed of cryptographic processes. 6 00:00:14.040 --> 00:00:15.930 Hardware acceleration is a 7 00:00:15.930 --> 00:00:18.630 primary performance consideration. 8 00:00:18.630 --> 00:00:22.830 Hardware acceleration uses specialized hardware components, 9 00:00:22.830 --> 00:00:25.470 such as cryptographic processors, 10 00:00:25.470 --> 00:00:27.090 or dedicated chips 11 00:00:27.090 --> 00:00:30.570 to perform encryption and decryption operations 12 00:00:30.570 --> 00:00:34.200 faster and more efficiently than would be possible 13 00:00:34.200 --> 00:00:35.910 with software alone. 14 00:00:35.910 --> 00:00:39.000 Let's learn more about hardware acceleration, 15 00:00:39.000 --> 00:00:42.300 then we'll conduct a quick demonstration. 16 00:00:42.300 --> 00:00:45.240 Hardware acceleration plays a critical role 17 00:00:45.240 --> 00:00:47.460 in balancing strong security 18 00:00:47.460 --> 00:00:49.170 with the need for efficient 19 00:00:49.170 --> 00:00:52.230 and speedy cryptographic processes. 20 00:00:52.230 --> 00:00:56.700 Hardware acceleration uses specialized hardware components 21 00:00:56.700 --> 00:00:58.950 like cryptographic processors, 22 00:00:58.950 --> 00:01:03.210 dedicated chips, and specialized hardware modules 23 00:01:03.210 --> 00:01:06.780 to perform encryption and decryption tasks 24 00:01:06.780 --> 00:01:10.800 much faster than standard software-based methods. 25 00:01:10.800 --> 00:01:13.920 Hardware acceleration component examples 26 00:01:13.920 --> 00:01:16.800 include hardware security modules, 27 00:01:16.800 --> 00:01:19.350 which manage cryptographic keys 28 00:01:19.350 --> 00:01:22.830 and perform encryption directly within the hardware, 29 00:01:22.830 --> 00:01:25.230 and trusted platform modules, 30 00:01:25.230 --> 00:01:29.010 which provide secure encryption, authentication, 31 00:01:29.010 --> 00:01:31.830 and key storage on computers. 32 00:01:31.830 --> 00:01:35.010 Graphic processing units, or GPUs, 33 00:01:35.010 --> 00:01:37.860 are also used for cryptographic tasks 34 00:01:37.860 --> 00:01:41.520 due to their parallel processing capabilities. 35 00:01:41.520 --> 00:01:45.090 Finally, network cards with built-in encryption, 36 00:01:45.090 --> 00:01:47.850 like Intel QuickAssist Technology, 37 00:01:47.850 --> 00:01:52.170 accelerate cryptographic functions at the network level. 38 00:01:52.170 --> 00:01:55.440 Each of these components are designed specifically 39 00:01:55.440 --> 00:01:58.620 to handle certain cryptographic algorithms, 40 00:01:58.620 --> 00:02:03.210 allowing them to carry out these operations more efficiently 41 00:02:03.210 --> 00:02:07.500 and reduce the load on general purpose CPUs. 42 00:02:07.500 --> 00:02:10.380 By offloading the cryptographic workload 43 00:02:10.380 --> 00:02:12.420 to these specialized chips, 44 00:02:12.420 --> 00:02:15.240 hardware acceleration significantly boosts 45 00:02:15.240 --> 00:02:17.670 overall system performance. 46 00:02:17.670 --> 00:02:19.650 So to get a better feel 47 00:02:19.650 --> 00:02:23.130 for how helpful hardware acceleration can be, 48 00:02:23.130 --> 00:02:25.950 imagine you are at a busy grocery store 49 00:02:25.950 --> 00:02:29.700 with a single cashier handling all the customers. 50 00:02:29.700 --> 00:02:30.990 The line moves slowly 51 00:02:30.990 --> 00:02:34.260 because the cashier has to scan each item, 52 00:02:34.260 --> 00:02:38.460 process payments, and bag groceries all on their own. 53 00:02:38.460 --> 00:02:40.410 Now, picture the store bringing in 54 00:02:40.410 --> 00:02:44.490 several other staff members to help the line move faster. 55 00:02:44.490 --> 00:02:48.330 One person scans the items, another handles payments, 56 00:02:48.330 --> 00:02:50.550 and a third bags the groceries. 57 00:02:50.550 --> 00:02:52.560 Suddenly, the checkout process 58 00:02:52.560 --> 00:02:55.500 becomes much faster and more efficient. 59 00:02:55.500 --> 00:02:59.370 This is like hardware acceleration in cryptography. 60 00:02:59.370 --> 00:03:02.970 Instead of relying on one general purpose component, 61 00:03:02.970 --> 00:03:05.520 like the cashier, to do everything, 62 00:03:05.520 --> 00:03:09.180 specialized hardware acts like the additional staff 63 00:03:09.180 --> 00:03:11.850 and takes on specific tasks. 64 00:03:11.850 --> 00:03:15.990 These components are optimized for speed and efficiency, 65 00:03:15.990 --> 00:03:19.800 allowing the system to handle encryption and decryption 66 00:03:19.800 --> 00:03:24.210 much faster, without overwhelming a single resource. 67 00:03:24.210 --> 00:03:28.020 Just as dividing tasks among specialized workers 68 00:03:28.020 --> 00:03:30.030 speeds up the checkout line, 69 00:03:30.030 --> 00:03:32.580 hardware acceleration boost the speed 70 00:03:32.580 --> 00:03:34.710 of cryptographic operations, 71 00:03:34.710 --> 00:03:38.640 making sensitives more efficient and responsive. 72 00:03:38.640 --> 00:03:41.370 This approach is especially valuable 73 00:03:41.370 --> 00:03:45.780 in environments where both speed and security are essential, 74 00:03:45.780 --> 00:03:49.320 such as high frequency trading platforms. 75 00:03:49.320 --> 00:03:52.980 In such scenarios, every millisecond counts, 76 00:03:52.980 --> 00:03:55.410 and hardware acceleration ensures 77 00:03:55.410 --> 00:03:59.790 that complex encryption tasks are executed rapidly 78 00:03:59.790 --> 00:04:04.140 without compromising the security of the transactions. 79 00:04:04.140 --> 00:04:06.840 By enhancing the speed of encryption, 80 00:04:06.840 --> 00:04:10.650 hardware acceleration helps maintain the high performance 81 00:04:10.650 --> 00:04:13.740 needed in these critical applications, 82 00:04:13.740 --> 00:04:16.230 allowing secure data processing 83 00:04:16.230 --> 00:04:19.110 without slowing down the system. 84 00:04:19.110 --> 00:04:21.630 In addition to trading platforms, 85 00:04:21.630 --> 00:04:26.010 hardware acceleration is also used in other applications, 86 00:04:26.010 --> 00:04:31.010 like virtual private networks and SSL/TLS protocols, 87 00:04:31.200 --> 00:04:34.320 to enhance performance and security. 88 00:04:34.320 --> 00:04:36.919 Secure sockets layer, or SSL, 89 00:04:36.919 --> 00:04:40.200 and TLS, or transport layer security, 90 00:04:40.200 --> 00:04:44.100 are protocols that establish secure communication channels 91 00:04:44.100 --> 00:04:47.040 between devices over the internet, 92 00:04:47.040 --> 00:04:49.620 such as when you visit a website 93 00:04:49.620 --> 00:04:52.920 or access a secure online service. 94 00:04:52.920 --> 00:04:56.430 These protocols rely heavily on encryption 95 00:04:56.430 --> 00:04:59.310 to protect the data being transmitted, 96 00:04:59.310 --> 00:05:03.660 ensuring that it remains private and tamper-proof. 97 00:05:03.660 --> 00:05:08.660 The encryption and decryption tasks involved in SSL and TLS 98 00:05:08.665 --> 00:05:11.760 are computationally intensive, 99 00:05:11.760 --> 00:05:15.090 especially when handling large volumes of data 100 00:05:15.090 --> 00:05:18.360 or many simultaneous connections. 101 00:05:18.360 --> 00:05:21.300 Hardware acceleration helps offload 102 00:05:21.300 --> 00:05:23.580 these cryptographic operations 103 00:05:23.580 --> 00:05:26.250 to specialized hardware components, 104 00:05:26.250 --> 00:05:30.810 such as cryptographic processors and dedicated chips, 105 00:05:30.810 --> 00:05:32.700 which can perform these tasks 106 00:05:32.700 --> 00:05:36.330 much faster than general purpose CPUs. 107 00:05:36.330 --> 00:05:38.640 This reduces the time required 108 00:05:38.640 --> 00:05:40.830 to establish secure connections 109 00:05:40.830 --> 00:05:43.050 and process encrypted data, 110 00:05:43.050 --> 00:05:46.680 making the overall system more efficient 111 00:05:46.680 --> 00:05:48.473 by using hardware acceleration. 112 00:05:48.473 --> 00:05:53.473 SSL and TLS can maintain high levels of security 113 00:05:53.483 --> 00:05:56.730 without slowing down network performance, 114 00:05:56.730 --> 00:05:59.370 ensuring that the encryption process 115 00:05:59.370 --> 00:06:01.770 does not become a bottleneck. 116 00:06:01.770 --> 00:06:03.990 This is particularly important 117 00:06:03.990 --> 00:06:06.300 in environments with high traffic, 118 00:06:06.300 --> 00:06:09.030 such as web servers, data centers, 119 00:06:09.030 --> 00:06:11.640 and online financial services, 120 00:06:11.640 --> 00:06:15.510 where fast and secure communication is essential. 121 00:06:15.510 --> 00:06:18.330 Let's do a quick demo to show the impact 122 00:06:18.330 --> 00:06:22.770 of hardware acceleration on cryptographic performance. 123 00:06:22.770 --> 00:06:27.770 I am on a Kali Linux machine with OpenSSL installed. 124 00:06:27.810 --> 00:06:31.320 OpenSSL includes a built-in speed test 125 00:06:31.320 --> 00:06:35.310 to measure the performance of cryptographic algorithms. 126 00:06:35.310 --> 00:06:38.430 This demonstration will show us the impact 127 00:06:38.430 --> 00:06:41.550 of removing hardware acceleration. 128 00:06:41.550 --> 00:06:45.300 We will start by running the OpenSSL speed test 129 00:06:45.300 --> 00:06:48.780 with hardware acceleration enabled by default. 130 00:06:48.780 --> 00:06:52.593 We'll do this with the following OpenSSL command. 131 00:06:57.990 --> 00:07:01.650 It'll take approximately 18 to 20 seconds 132 00:07:01.650 --> 00:07:04.260 for this command to execute. 133 00:07:04.260 --> 00:07:07.410 It is measuring the performance of AES 134 00:07:07.410 --> 00:07:12.410 with a 256-bit key in Cipher Block Chaining, or CBC, mode. 135 00:07:14.250 --> 00:07:18.600 The output is showing us the number of operations per second 136 00:07:18.600 --> 00:07:21.600 that are completed with different block sizes. 137 00:07:21.600 --> 00:07:24.690 You can see that we have 16-byte blocks, 138 00:07:24.690 --> 00:07:27.630 64-byte blocks, et cetera, 139 00:07:27.630 --> 00:07:30.990 and each one of the numbers that is provided here 140 00:07:30.990 --> 00:07:32.100 as the result 141 00:07:32.100 --> 00:07:35.610 are in thousands of bytes per second processed. 142 00:07:35.610 --> 00:07:38.310 Let's just take a look at the overall length 143 00:07:38.310 --> 00:07:41.160 of each one of these results. 144 00:07:41.160 --> 00:07:43.470 In the 16-byte block, 145 00:07:43.470 --> 00:07:48.000 we can see that the output we receive is seven digits long. 146 00:07:48.000 --> 00:07:50.700 That's true for each one of these blocks, 147 00:07:50.700 --> 00:07:52.560 meaning that we're processing, 148 00:07:52.560 --> 00:07:57.180 let's say in the 16384 byte block size, 149 00:07:57.180 --> 00:08:02.180 1,410,088 thousands of bytes per second 150 00:08:03.510 --> 00:08:05.280 that have been processed. 151 00:08:05.280 --> 00:08:09.090 Okay, now let's remove hardware acceleration 152 00:08:09.090 --> 00:08:14.040 with the following export OPENSSL command. 153 00:08:14.040 --> 00:08:16.920 Now that hardware acceleration is removed, 154 00:08:16.920 --> 00:08:19.710 we're going to rerun that speed test 155 00:08:19.710 --> 00:08:22.500 that we ran the very first time. 156 00:08:22.500 --> 00:08:25.140 Again, what we're going to find is it takes 157 00:08:25.140 --> 00:08:30.140 approximately 18 to 20 seconds for this command to execute. 158 00:08:30.990 --> 00:08:34.110 At the end, what we expect to see 159 00:08:34.110 --> 00:08:36.660 with hardware acceleration removed 160 00:08:36.660 --> 00:08:40.050 is that the output numbers that we receive at the bottom 161 00:08:40.050 --> 00:08:44.400 are much lower than the original test that we ran, 162 00:08:44.400 --> 00:08:47.340 and we can do that by just looking at the length 163 00:08:47.340 --> 00:08:49.290 of each one of these outputs. 164 00:08:49.290 --> 00:08:53.070 Instead of the outputs being in the millions, 165 00:08:53.070 --> 00:08:57.450 we're now at 271,180 166 00:08:57.450 --> 00:08:59.640 for the 16-byte block, 167 00:08:59.640 --> 00:09:04.640 all the way to 283,262 168 00:09:04.860 --> 00:09:08.700 in the 16384 byte block. 169 00:09:08.700 --> 00:09:12.480 So as you can see, the number of bytes per second 170 00:09:12.480 --> 00:09:13.950 that have been processed 171 00:09:13.950 --> 00:09:16.620 is no longer in the millions 172 00:09:16.620 --> 00:09:18.690 for each one of these block sizes. 173 00:09:18.690 --> 00:09:21.240 It's now in the hundreds of thousands 174 00:09:21.240 --> 00:09:23.370 for each one of these block sizes, 175 00:09:23.370 --> 00:09:26.190 showing a significant decrease 176 00:09:26.190 --> 00:09:30.000 in the number of bytes per second that are processed 177 00:09:30.000 --> 00:09:33.270 when hardware acceleration is removed. 178 00:09:33.270 --> 00:09:37.170 So remember, performance considerations 179 00:09:37.170 --> 00:09:40.020 balance the need for strong security 180 00:09:40.020 --> 00:09:44.310 with the speed and efficiency of cryptographic processes. 181 00:09:44.310 --> 00:09:47.400 Hardware acceleration plays a key role 182 00:09:47.400 --> 00:09:49.470 in enhancing these processes 183 00:09:49.470 --> 00:09:51.960 by using specialized components, 184 00:09:51.960 --> 00:09:56.310 such as cryptographic processors and dedicated chips, 185 00:09:56.310 --> 00:09:59.130 to perform the encryption and decryption 186 00:09:59.130 --> 00:10:02.520 much faster than software could alone. 187 00:10:02.520 --> 00:10:04.440 These hardware components, 188 00:10:04.440 --> 00:10:09.000 like hardware security modules and trusted platform modules, 189 00:10:09.000 --> 00:10:10.740 are specifically designed 190 00:10:10.740 --> 00:10:13.920 to handle certain cryptographic algorithms, 191 00:10:13.920 --> 00:10:17.550 reducing the load on general purpose CPUs 192 00:10:17.550 --> 00:10:20.820 and boosting overall system performance. 193 00:10:20.820 --> 00:10:24.030 This approach is particularly valuable 194 00:10:24.030 --> 00:10:28.440 in environments where both speed and security are critical, 195 00:10:28.440 --> 00:10:33.440 as it ensures that encryption tasks are executed rapidly 196 00:10:33.660 --> 00:10:36.330 without compromising security. 197 00:10:36.330 --> 00:10:39.030 By offloading cryptographic workloads 198 00:10:39.030 --> 00:10:40.920 to specialized hardware, 199 00:10:40.920 --> 00:10:43.080 hardware acceleration enables 200 00:10:43.080 --> 00:10:46.080 secure and efficient data processing 201 00:10:46.080 --> 00:10:48.450 in high-demand applications, 202 00:10:48.450 --> 00:10:53.450 like virtual private networks and SSL or TLS protocols, 203 00:10:54.150 --> 00:10:56.640 maintaining excellent performance, 204 00:10:56.640 --> 00:10:58.833 even under heavy loads.