ENSA Module 13 Network Virtualization — Transcript

Full transcript

1. 0:03[Music]
2. 0:12hi
3. 0:12hello there welcome to network
4. 0:15virtualization
5. 0:17so imagine you live in a two-bedroom
6. 0:19house
7. 0:20you use the second bedroom for storage
8. 0:24the second bedroom is packed full of
9. 0:26boxes
10. 0:27but you still have more to place in
11. 0:30storage
12. 0:31so you should consider building an
13. 0:32addition on your house
14. 0:35it would be costly endeavor and you may
15. 0:38not need that much space forever
16. 0:40so you decide to rent a storage unit
17. 0:44for the overflow so similar to storage
18. 0:47unit
19. 0:47network virtualization and cloud
20. 0:50services
21. 0:51can provide a business with options
22. 0:53other than adding servers
23. 0:55into their own data center in addition
24. 0:58to storage
25. 0:59it offers other advantages so get
26. 1:02started with this module
27. 1:03to learn more about what virtualization
28. 1:06and cloud services can do
29. 1:13so on this video lecture we will be
30. 1:15talking about network virtualization
31. 1:17and the module objective is to explain
32. 1:19the purpose and characteristics of
33. 1:21network virtualization
34. 1:23so along with this we will be talking
35. 1:25about the cloud computing
36. 1:27virtualization the virtual network
37. 1:29infrastructure
38. 1:30the software defined networking and the
39. 1:32controllers
40. 1:37so let's start with cloud computing
41. 1:40so cloud computing involves large
42. 1:43numbers of computers connected to a
43. 1:45network
44. 1:46that can be physically located anywhere
45. 1:49so providers
46. 1:50rely heavily on virtualization to
47. 1:52deliver their cloud
48. 1:54computing services so cloud computing
49. 1:56services can reduce operational costs
50. 1:59by using resources more efficiently so
51. 2:02cloud computing
52. 2:03addresses a variety of data management
53. 2:05issues
54. 2:07like it enables access to organizational
55. 2:10data
56. 2:11anywhere and at any time
57. 2:14next is it streamlines the
58. 2:16organization's id operations
59. 2:18by subscribing only to needed services
60. 2:22okay next is it eliminates or reduces
61. 2:26the need for on-site id equipment
62. 2:28maintenance and management
63. 2:32also it reduces cost of equipment
64. 2:35energy physical plant requirements and
65. 2:39personal training needs and the last one
66. 2:42is
67. 2:42it enables rapid responses to
68. 2:45increasing data volume requirements so
69. 2:48cloud computing
70. 2:49with its pay-as-you-go model allows
71. 2:52organization to trip
72. 2:54computing and storage expenses more as
73. 2:57utility
74. 2:58rather than investing in the
75. 2:59infrastructure so capital expenditures
76. 3:03are transformed into operating
77. 3:05expenditures
78. 3:08okay so let's talk about cloud services
79. 3:10now
80. 3:13so cloud services are available in a
81. 3:15variety of options
82. 3:16tailored to meet customer requirements
83. 3:19the three main cloud computing services
84. 3:21defined by the national institute of
85. 3:22standards in technology or nist
86. 3:25in their special publication 800.145
87. 3:29are as follows okay so the first one is
88. 3:33software as a service or sas okay
89. 3:37so the cloud provider is responsible for
90. 3:39access to applications and services such
91. 3:41as email
92. 3:42communication and office 365
93. 3:46that are delivered over the internet so
94. 3:48the user does not manage
95. 3:49any aspect of the cloud services except
96. 3:52for limited
97. 3:53user specific application settings so
98. 3:56the user only needs to provide
99. 3:58their data okay so the next one is
100. 4:02platform as a service or us
101. 4:05so the cloud provider is responsible for
102. 4:07providing user access to the development
103. 4:09tools and services
104. 4:11used to deliver the applications the
105. 4:14users are typically programmers
106. 4:16and may have control over the
107. 4:18configuration settings
108. 4:20of the cloud provider's application
109. 4:22hosting environment
110. 4:25next is infrastructure as a service
111. 4:28okay infrastructure as a service or eos
112. 4:32the cloud provider is responsible for
113. 4:34giving it managers
114. 4:36access to network equipment virtualized
115. 4:39network services
116. 4:40and supporting network infrastructure so
117. 4:43using this cloud service allows it
118. 4:46manager
119. 4:47to deploy and run software code which
120. 4:49can include operating systems and
121. 4:51applications
122. 4:53all right so cloud service providers
123. 4:56have extended this model
124. 4:59to also provide an i.t support to each
125. 5:01of the cloud computing services
126. 5:02and what we have here is of course the
127. 5:04it as a service
128. 5:06or itas okay so for businesses
129. 5:09itas can extend the capability of
130. 5:12network without requiring investment in
131. 5:14new infrastructure
132. 5:15training new personnel or licensing new
133. 5:18software
134. 5:19so these services are available on
135. 5:21demand and delivered economically
136. 5:23to any device anywhere in the world
137. 5:25without compromising security or
138. 5:27function
139. 5:37okay next is cloud models
140. 5:41okay so there are four cloud models as
141. 5:44shown in the figure here
142. 5:46you've got hybrid public private and
143. 5:49community
144. 5:50okay so public clouds okay let's start
145. 5:53with public clouds
146. 5:54public clouds are cloud-based
147. 5:56applications and services
148. 5:58offered in public cloud that are made
149. 6:00available
150. 6:01to the general population so services
151. 6:04may be free or
152. 6:06are offered on a pay-per-use model such
153. 6:09as paying for online storage
154. 6:11okay so the public cloud uses the
155. 6:14internet to provide services
156. 6:16so the next one is private cloud okay
157. 6:20so cloud-based applications and services
158. 6:22offered in private cloud
159. 6:24are intended for specific organization
160. 6:26or entity
161. 6:28such as the government your private
162. 6:30cloud can set up
163. 6:31using the organization's private network
164. 6:34so through this can be expensive to
165. 6:36build
166. 6:37okay and maintain so a private cloud can
167. 6:41also be managed by an outside
168. 6:43organization
169. 6:45with strict access security
170. 6:48next third one is the hybrid hybrid
171. 6:51cloud
172. 6:52okay so a hybrid cloud is made up of two
173. 6:55or more clouds example
174. 6:57part private and part public so
175. 7:00where each part remains a separate
176. 7:03object
177. 7:04but both are connected using a single
178. 7:07architecture
179. 7:08so individuals on a hybrid cloud would
180. 7:10be able to decree or to have
181. 7:12degrees of access okay so to vary
182. 7:15services
183. 7:16based on the user access rights all
184. 7:19right
185. 7:19and the last one is the community clouds
186. 7:22so a community cloud is created for
187. 7:25exclusive use
188. 7:26by a specific community so the
189. 7:28differences between
190. 7:30public clouds and community clouds are
191. 7:32the functional needs
192. 7:34that have been customized for the
193. 7:35community for example
194. 7:38healthcare okay healthcare organizations
195. 7:41must remain compliant with policies and
196. 7:44laws example the
197. 7:45hipaa that requires special
198. 7:48authentication and confidentiality
199. 7:53right next cloud computing versus data
200. 7:57center
201. 7:58okay so the term data center and cloud
202. 8:01computing are often used incorrectly
203. 8:04okay so these are the correct
204. 8:05definitions of data center and cloud
205. 8:07computing
206. 8:09so data center okay so typically
207. 8:13an off premise or a data storage
208. 8:17and processing facility run by an
209. 8:19in-house id department or list
210. 8:21offsite so that is a data center when
211. 8:24you say cloud computing so typically
212. 8:27this is an off-premise service that
213. 8:30offers an
214. 8:31on-demand access to a shared pool of
215. 8:34configurable
216. 8:35computing resources so these resources
217. 8:38can be
218. 8:38rapidly provisioned and released with
219. 8:41minimal management effort
220. 8:43so data centers are the physical
221. 8:45facilities that provide the compute
222. 8:47network and storage needs of cloud
223. 8:50computing services
224. 8:52while cloud service providers use data
225. 8:55centers
226. 8:56to host their cloud services and
227. 8:59cloud-based resources
228. 9:01all right
229. 9:05let's talk about virtualization now
230. 9:08okay so cloud computing and
231. 9:11virtualization
232. 9:12what's the difference so in the previous
233. 9:15topic you learned about cloud services
234. 9:17and cloud models
235. 9:19so this topic will explain
236. 9:21virtualization
237. 9:23so the terms cloud computing and
238. 9:25virtualization
239. 9:26are often used interchangeably however
240. 9:30they mean different things all right
241. 9:33so virtualization is the foundation of
242. 9:35cloud computing
243. 9:36without it cloud computing as it is most
244. 9:40widely
245. 9:40implemented would not be possible so
246. 9:44virtualization separates the operating
247. 9:46systems or os
248. 9:47from the hardware so various providers
249. 9:50offer virtual cloud services that can
250. 9:54dynamically provision servers
251. 9:56as required for example amazon web
252. 9:58services or aws
253. 10:00provides a simple way for customers to
254. 10:04dynamically provision
255. 10:05the compute resources they need these
256. 10:08virtualized instances of servers
257. 10:11are created on demand okay
258. 10:14so as shown in the figure here all right
259. 10:18so the network administrator can deploy
260. 10:20a variety of services from the aws
261. 10:22management console
262. 10:24including virtual machines web
263. 10:26applications
264. 10:27virtual servers and connections to iot
265. 10:30devices
266. 10:32all right okay let's talk about
267. 10:35dedicated servers now
268. 10:36so to fully appreciate virtualization it
269. 10:39is first necessary to understand
270. 10:41some of the history of the server
271. 10:43technology
272. 10:44so historically enterprise servers
273. 10:47consisted
274. 10:48of a server os such as windows server or
275. 10:50linux server
276. 10:52okay installed on the specific hardware
277. 10:54as shown in the figure
278. 10:56so all of the servers ram processing
279. 10:58power
280. 10:59and hard disk drive spaces were
281. 11:02dedicated to the service provided
282. 11:04example web email or email services
283. 11:08and others so the major problem with
284. 11:10this configuration is that
285. 11:12when a component fails the service that
286. 11:15is provided by that server becomes
287. 11:17unavailable
288. 11:18so this is known as a single point of
289. 11:20failure
290. 11:21another problem was that dedicated
291. 11:24servers
292. 11:25were underused right so dedicated
293. 11:28servers
294. 11:28can often sat idle for long periods of
295. 11:31time
296. 11:32so waiting until there was a need to
297. 11:34deliver the specific service that they
298. 11:36provide
299. 11:37so this servers wasted energy and took
300. 11:40up
301. 11:40more space than was warranted by the
302. 11:44amount of service provided
303. 11:47this is known as the server sprawl
304. 11:55all right so next is virtualization
305. 11:58or server virtualization so server
306. 12:01virtualization takes advantage of
307. 12:04idle resources and consolidates the
308. 12:06number of required servers
309. 12:08so this also allows for multiple
310. 12:10operating systems
311. 12:12to exist on a single hardware platform
312. 12:14for example
313. 12:15in the figure the previous eight
314. 12:17dedicated servers
315. 12:19have been consolidated into two servers
316. 12:21using hypervisors
317. 12:23to support multiple virtual instances of
318. 12:25the operating systems
319. 12:29the use of virtualization normally
320. 12:31includes redundancy to protect
321. 12:33from a single point of failure so
322. 12:35redundancy can be implemented in
323. 12:37different ways
324. 12:38if the high fare buys or fails the
325. 12:41virtual machine can be restarted on
326. 12:43another hypervisor also the same vm
327. 12:47can run on two hypervisors concurrently
328. 12:50so copying the ram and cpu instructions
329. 12:53between them
330. 12:55so if one hypervisor fails the virtual
331. 12:58machine continues
332. 13:00running on another hypervisor
333. 13:03okay so the services running on the vms
334. 13:06are also virtual
335. 13:08and can be dynamically installed or
336. 13:10uninstalled as needed
337. 13:12so the hypervisor is a program or
338. 13:14firmware
339. 13:15or hardware that adds an abstraction
340. 13:17layer
341. 13:18okay on top of the physical hardware
342. 13:21so the abstraction layer is used to
343. 13:23create a virtual machines
344. 13:25which have access to all the hardware of
345. 13:28the physical machines such as cpus
346. 13:30memory disk controllers and network
347. 13:33interface cards
348. 13:36so each of these virtual machines runs
349. 13:38on a consolidated number of servers they
350. 13:41require
351. 13:41for example it is not uncommon
352. 13:45for 100 physical servers to be
353. 13:47consolidated as virtual machines on top
354. 13:49of
355. 13:5010 physical servers that are using
356. 13:52hypervisors
357. 13:54okay so let's talk about the advantages
358. 13:56of virtualization now
359. 13:58so one major advantage of virtualization
360. 14:01is
361. 14:02overall reduced cost so you've got less
362. 14:05equipment is required
363. 14:06less energy is consumed and less space
364. 14:10is required okay now
365. 14:13with less equipment is required
366. 14:15virtualization enables server
367. 14:17consolidation
368. 14:18which requires fewer physical servers
369. 14:21fewer networking devices
370. 14:23and less supporting infrastructure so it
371. 14:25means lower maintenance
372. 14:28cost so the next one is less energy is
373. 14:31consumed
374. 14:32so consolidating servers lowers the
375. 14:34monthly power and cooling costs
376. 14:37so reduced consumption helps enterprises
377. 14:39to achieve a smaller carbon footprint
378. 14:42okay when you say less space is required
379. 14:46server consolidation with virtualization
380. 14:49reduces the overall footprint of the
381. 14:51data center
382. 14:52so fewer servers network devices
383. 14:56and racks reduce the amount of the
384. 14:57required floor spaces
385. 15:00okay so there are additional benefits of
386. 15:03virtualization so like
387. 15:05easier prototyping so self-contained
388. 15:08labs operating on isolated networks can
389. 15:12be rapidly created for testing and
390. 15:13prototyping network deployments
391. 15:16so if a mistake is made an administrator
392. 15:19can simply revert
393. 15:20their previous version so the testing
394. 15:23environment can be online
395. 15:25but isolate that from the end users so
396. 15:28when testing is completed
397. 15:29the servers and systems can be deployed
398. 15:32to the end servers and end users
399. 15:36okay the next one is faster server
400. 15:39provisioning
401. 15:40so creating a virtual server is far
402. 15:42faster than provisioning a physical
403. 15:44server
404. 15:45okay next is increased server uptime
405. 15:49most server utilization or
406. 15:51virtualization
407. 15:53platforms now offer advanced redundant
408. 15:55fault tolerance of features such as
409. 15:57live migration storage migration high
410. 16:01availability
411. 16:02and distributed resource scheduling
412. 16:05okay so next additional benefits
413. 16:08would be improved disaster recovery
414. 16:12so virtualization offers advanced
415. 16:14business continuity solutions
416. 16:16so it provides hardware abstraction
417. 16:18capability
418. 16:20so that the recovery site no longer
419. 16:22needs to have hardware
420. 16:23that is identical to the hardware in
421. 16:25their production environment
422. 16:27so most enterprises or enterprise server
423. 16:30virtualization platform
424. 16:32also have software that can help test
425. 16:34and automate
426. 16:35the failover over or before a disaster
427. 16:39does happen okay so the last one is
428. 16:43legacy support so virtualization can
429. 16:45extend the life of oss
430. 16:47and application providing more time for
431. 16:50organization
432. 16:51to migrate to newer solutions
433. 16:55okay
434. 16:58now let's talk about obstruction layers
435. 17:01okay so to help explain
436. 17:03how virtualization works it helps to use
437. 17:06layers of obstruction in computer
438. 17:08architecture
439. 17:09the computer system consists of the
440. 17:11following abstraction layers as
441. 17:13illustrated here
442. 17:14so you've got hardware which includes
443. 17:16cpu memory and ic and disk or hard disk
444. 17:19storage
445. 17:20you've got firmware or rom okay the
446. 17:23operating systems are os
447. 17:25and the services like web email and
448. 17:28file service okay so at each of these
449. 17:32layers of abstraction
450. 17:34some type of programming code is used
451. 17:37as an interface between the layer below
452. 17:40the layer above
453. 17:42okay so that is between the layer below
454. 17:45and the layer above so for example
455. 17:47the c programming language is often used
456. 17:49to program the firmware
457. 17:51that accesses the hardware so an example
458. 17:54of virtualization is shown in the figure
459. 17:57okay so a hypervisor is installed
460. 18:00between the firmware
461. 18:01okay and the os
462. 18:05okay so the hypervisor can support
463. 18:07multiple instances of oss
464. 18:10like what you see here in the diagram
465. 18:12okay so this hypervisor
466. 18:15has three oss running on top of it so
467. 18:18that means
468. 18:19we could have multiple oss running and
469. 18:23its os could run its own service
470. 18:26all right
471. 18:30next let's talk about type 2 hypervisors
472. 18:34so a type to hypervisor is software
473. 18:37that creates and runs virtual machine
474. 18:39instances
475. 18:41so the computer on which a hypervisor is
476. 18:43supporting one or more vms
477. 18:45is a host machine so type 2 hypervisors
478. 18:48are also called hosted
479. 18:50hypervisors this is because the
480. 18:53hypervisor
481. 18:54is installed on top of the existing
482. 18:56operating systems such as mac os
483. 18:59windows or linux then
484. 19:02one or more additional os instances are
485. 19:04installed on top of the hypervisor
486. 19:06as shown here in the diagram okay so for
487. 19:09instance
488. 19:10we have here a single host okay so
489. 19:12definitely we have the shared resources
490. 19:14or hardware
491. 19:15and then we have what you call the host
492. 19:17operating system
493. 19:19okay on top of this host operating
494. 19:21system we have a hypervisor running
495. 19:24which is hosting different oss on top of
496. 19:27it
497. 19:28now in this example we are running three
498. 19:30virtual machines
499. 19:31powered by windows os linux os and
500. 19:34the unix oss okay so the big advantage
501. 19:38of a type to hypervisor is that
502. 19:40management console software is not
503. 19:43required
504. 19:44okay so the type to hypervisor are very
505. 19:47popular with
506. 19:48consumers for organizations
507. 19:50experimenting with virtualization
508. 19:53so common type to hypervisor includes
509. 19:55virtual pc
510. 19:57or the vmware workstation you've got the
511. 19:59oracle virtual box
512. 20:02vmware fusion and the mac os x
513. 20:05parallels okay so many of these type two
514. 20:09hypervisors are free
515. 20:10however some hypervisors offer more
516. 20:13advanced features for a fee
517. 20:16okay so note that it is important to
518. 20:19make sure that the host machine is
519. 20:20robust
520. 20:21enough to install and run the virtual
521. 20:23machines
522. 20:24so that it does not run out of the
523. 20:27resources
524. 20:28because everything is shared when you
525. 20:30run
526. 20:31and use a type to hypervisors
527. 20:38okay so let's talk about virtual network
528. 20:41infrastructure
529. 20:43okay so under virtual network
530. 20:46infrastructure we've got the type
531. 20:481 hypervisor so in the previous topic
532. 20:51you learned about virtualization
533. 20:53this topic will cover the virtual
534. 20:55network infrastructure
535. 20:58all right so type 1 hypervisors are also
536. 21:01called bare metal approach
537. 21:03because the hypervisor is installed
538. 21:05directly on the hardware
539. 21:07so type 1 hypervisors are usually used
540. 21:10on enterprise servers and data center
541. 21:12networking devices
542. 21:14so with type 1 hypervisor the hypervisor
543. 21:18is installed directly
544. 21:19on the server or network hardware then
545. 21:22instances
546. 21:23of an operating systems are installed on
547. 21:25the hypervisor
548. 21:26as on in the diagram here okay so we
549. 21:30have the server
550. 21:31the hardware hypervisor oss
551. 21:34are installed directly on the hypervisor
552. 21:36so we don't have a host os here
553. 21:38on the physical device all right
554. 21:42so we type 1 hypervisor the hypervisor
555. 21:44is installed directly on the server
556. 21:46or networking hardware then instances of
557. 21:49an os are installed on the hypervisor
558. 21:52as i shown you earlier okay
559. 21:55so type 1 hypervisor have direct access
560. 21:58to the hardware resources
561. 22:00therefore they are more efficient than
562. 22:03hosted architectures
563. 22:05so type 1 hypervisors improve
564. 22:08scalability
565. 22:09performance and robustness
566. 22:15how to install a virtual machine on a
567. 22:17hypervisor
568. 22:18so when a type 1 hypervisor is installed
569. 22:21and the server is rebooted
570. 22:23only basic information is displayed such
571. 22:25as the os version
572. 22:27the amount of rom and the ip addresses
573. 22:30so an os instance cannot be created from
574. 22:33this screen
575. 22:34type 1 hypervisors require a management
576. 22:37console
577. 22:38to manage the hypervisor so management
578. 22:41software is used to manage multiple
579. 22:43servers
580. 22:44using the same hypervisor so the
581. 22:46management console can automatically
582. 22:48consolidate servers
583. 22:50and power on and off servers as required
584. 22:57okay so for example assume that server
585. 23:00one
586. 23:01in the figure okay becomes law on
587. 23:05resources
588. 23:06so to make more resources available the
589. 23:08network administrator
590. 23:10uses the management console
591. 23:13to move the windows instance to the
592. 23:15hypervisor on server too
593. 23:18all right so the management console can
594. 23:21also be programmed
595. 23:22with thresholds that will trigger more
596. 23:25automatically
597. 23:27okay so it could also trigger the
598. 23:29movement
599. 23:30of this windows os automatically
600. 23:34okay now the figure here
601. 23:37below okay the management console
602. 23:40provides recovery from hardware failure
603. 23:42so if a server component fails the
604. 23:45management console
605. 23:46automatically moves the virtual machine
606. 23:48to another server
607. 23:50so the management console for the cisco
608. 23:52unified computing system or ucs manager
609. 23:55is shown here in the figure so the cisco
610. 23:58ucs manager controls
611. 23:59multiple servers and manages resources
612. 24:01for thousands of
613. 24:03virtual machines so some management
614. 24:05consoles also allow server over
615. 24:08allocation over allocation is
616. 24:11when multiple os instances are installed
617. 24:13but their memory allocation exceeds
618. 24:16the total amount of memory that the
619. 24:18server has for example
620. 24:20a server has 16 gb of ram but the
621. 24:23administrator creates
622. 24:25four os instances with 10 gb allocated
623. 24:28to it
624. 24:29so this type of overall allocation is a
625. 24:32common practice because
626. 24:34all four os instances rarely require
627. 24:37the full 10 gb of ram at
628. 24:40any one moment okay
629. 24:48the complexity of network virtualization
630. 24:51okay so how complex virtualization or
631. 24:54network virtualization is
632. 24:56so server virtualization hides server
633. 24:59resources
634. 25:00such as the number and identity of
635. 25:02physical servers processors
636. 25:04and oss from server users so this
637. 25:08practice can create
638. 25:09problems if the data center is using
639. 25:11traditional network architectures for
640. 25:14example
641. 25:15virtual lands or vlans used by vms must
642. 25:18be assigned to the same
643. 25:19switch port as the physical server
644. 25:22running on the hypervisor
645. 25:23however vms are movable
646. 25:27and the network administrator must be
647. 25:29able to add
648. 25:30drop or change network resources and
649. 25:32profiles
650. 25:33so this process would be manual and time
651. 25:36consuming
652. 25:36with additional network switches so
653. 25:39another problem
654. 25:40is that traffic flows differ
655. 25:42substantially from
656. 25:44the traditional client server model so
657. 25:46typically
658. 25:47a data center has a considerable amount
659. 25:50of traffic being exchanged between
660. 25:52virtual servers such as
661. 25:53the ucl servers okay as shown here in
662. 25:57the diagram
663. 26:00all right so this uses a ucs server
664. 26:04now these flows are called east west
665. 26:07traffic
666. 26:08okay and can change in location and
667. 26:11intensity over time
668. 26:13so the north south traffic of course
669. 26:15between the distribution and the core
670. 26:16layers
671. 26:18all right so you've got the north south
672. 26:20traffic between the core
673. 26:21and then the distribution layers okay
674. 26:24so north south traffic occurs between
675. 26:26the distribution and the core layer and
676. 26:28is typically
677. 26:29traffic distant for off-site locations
678. 26:32such as another data center
679. 26:34other cloud providers or the internet
680. 26:38right so the east-west traffic is
681. 26:41within the ucs servers that is connected
682. 26:43to the access layer
683. 26:44switches
684. 26:49next so let's talk about the complexity
685. 26:51of network virtualization
686. 26:53okay on dynamic ever-changing traffic
687. 26:56requires a flexible approach
688. 26:58on network resource management okay so
689. 27:01existing network
690. 27:02infrastructure can respond to changing
691. 27:05requirements related to the management
692. 27:07of traffic flows by using the quality of
693. 27:09service or qos
694. 27:11and security level configurations for
695. 27:13individual flaws
696. 27:15however in large enterprises using
697. 27:18multi-vendor equipment
698. 27:20each time a new virtual machine is
699. 27:22enabled
700. 27:23the necessary reconfiguration can be
701. 27:26very time consuming
702. 27:28so the network infrastructure can also
703. 27:30benefit from virtualization
704. 27:32so network functions can be virtualized
705. 27:35each network device
706. 27:36can be segmented into multiple virtual
707. 27:39devices
708. 27:40that operate as independent devices
709. 27:43examples includes sub-interfaces
710. 27:46virtual interfaces vlans and routing
711. 27:49tables
712. 27:50so virtualized routing is called virtual
713. 27:53routing
714. 27:54and forwarding or vrf
715. 27:58okay now how is the network virtualized
716. 28:02the answer is found on how a networking
717. 28:05device
718. 28:06operates using a data plane and control
719. 28:08pane
720. 28:09as discussed on the next topic
721. 28:14let's talk about the software defined
722. 28:15networking now
723. 28:19the previous topic explained virtual
724. 28:21network infrastructure
725. 28:22this topic will cover the software
726. 28:24defined networking
727. 28:27so a network device contains the
728. 28:29following planes you've got the control
729. 28:31plane
730. 28:31and the data plane now the control plane
731. 28:36this is typically regarded as the brains
732. 28:38of the device
733. 28:39so it is used to make forwarding
734. 28:41decisions the control plane
735. 28:44contains layer 2 and layer 3 route
736. 28:46forwarding mechanisms
737. 28:48such as routing protocol neighbor tables
738. 28:51and topology tables ipb4 and ipv6
739. 28:54routing tables
740. 28:56stp and the arp table
741. 28:59so information sent to the control plane
742. 29:02is processed by
743. 29:03the cpu now on the data plane
744. 29:07also called the forwarding plane this
745. 29:10plane is typically in the switch fabric
746. 29:12connecting the various network ports on
747. 29:15a device
748. 29:16so the data plane of each device is used
749. 29:19to forward traffic flows
750. 29:21routers and switches use information
751. 29:23from the control plane
752. 29:25to forward incoming traffic out of the
753. 29:28appropriate egress interfaces
754. 29:30so information in the data plane is
755. 29:32typically processed by special data
756. 29:34plane processor
757. 29:36without the cpu getting involved
758. 29:42now illustration and explanation of the
759. 29:44difference between the operation of
760. 29:46localized
761. 29:46control on a layer 3 switch and a
762. 29:49centralized controller is the end
763. 29:51follows okay so
764. 29:55beginning with the layer 3 switch and
765. 29:58ceph or the cisco express forwarding
766. 30:00okay now this figure here illustrates
767. 30:04how cisco
768. 30:05express forwarding or cf uses the
769. 30:07control plane
770. 30:08and data plane to process packets
771. 30:11you've got the control plane on top and
772. 30:13the data plane at the bottom
773. 30:15okay egress interface that is where the
774. 30:18packet is coming in
775. 30:19and egress interface would be the output
776. 30:22or the
777. 30:23packets going out okay now ceph
778. 30:26is an advanced layer 3 ip switching
779. 30:29technology
780. 30:30that enables forwarding to packets or of
781. 30:32packets to occur
782. 30:34at the data plane without consulting the
783. 30:36control pane
784. 30:37in ceph the control plane's routing
785. 30:40table
786. 30:41pre-populates the cef forwarding
787. 30:43information base or fib table
788. 30:45in the data plane so the control plane's
789. 30:48arp
790. 30:49table pre-populates the adjacency table
791. 30:52so packets are then forwarded directly
792. 30:55by the data plane
793. 30:56based on the information contained in
794. 30:58the fib and adjacency table
795. 31:01without needing to consult information
796. 31:04in the control plane
797. 31:08okay so the next one is the software
798. 31:11defined networking
799. 31:13okay so which is basically the
800. 31:14separation of the control plane and data
801. 31:16plane
802. 31:17the control plane function is removed
803. 31:19from each device and is performed by a
804. 31:22centralized controller
805. 31:23as shown in the figure here okay so if
806. 31:26you will observe here
807. 31:28if you have different devices what is
808. 31:30left
809. 31:31on the device is just the data plane now
810. 31:34the control plane was removed
811. 31:35from this device okay and is
812. 31:38placed on a centralized controller okay
813. 31:42now the centralized controller
814. 31:43communicates control plane functions to
815. 31:45each device
816. 31:47each device can now focus on forwarding
817. 31:49data
818. 31:50while the centralized controller manages
819. 31:52data flow increases security and
820. 31:54provides other services
821. 32:00so the management plane is the plane
822. 32:03which is responsible for managing a
823. 32:05device
824. 32:06through its connection to the network
825. 32:08network administrators and systems
826. 32:10administrators
827. 32:11use the application such as secure shell
828. 32:14or ssh
829. 32:15the trivial file transfer protocol or
830. 32:17tftp
831. 32:19the secure ftp and the secure hyper text
832. 32:22transfer protocol or
833. 32:24https to access the management plane and
834. 32:27configure a device
835. 32:28so the management plane is how you have
836. 32:31accessed and configured devices in your
837. 32:34networking studies
838. 32:36in addition protocols like simple
839. 32:39network management protocol or snmp
840. 32:42is used in the management plan
841. 32:49so there are two major network
842. 32:51architectures that have been developed
843. 32:53to support the network virtualization
844. 32:55okay
845. 32:56so over a decade ago vmware developed a
846. 32:59virtualizing technology
847. 33:01that enabled a host os to support one or
848. 33:04more of client voices
849. 33:06most virtualization technologies are now
850. 33:08based on this technology
851. 33:10the transformation of dedicated servers
852. 33:12to virtualized servers
853. 33:14has or has been embraced and is rapidly
854. 33:17being implemented
855. 33:18in data centers and enterprise networks
856. 33:21okay now the two major network
857. 33:25architectures
858. 33:26are the software defined networking for
859. 33:28sdn
860. 33:30and the cisco application centric
861. 33:31infrastructure or the aci
862. 33:34now sdn is a network architecture that
863. 33:37virtualizes the network
864. 33:39offering a new approach to network
865. 33:41administration
866. 33:42and management that seeks to simplify
867. 33:45and streamline the administration
868. 33:46process
869. 33:48now the cisco application centric
870. 33:50infrastructure or aci
871. 33:53a purpose built hardware solution
872. 33:57for integrating cloud computing and data
873. 33:59center management
874. 34:05now here are the components of the sdn
875. 34:08okay so with it which includes an open
876. 34:10flow
877. 34:10and open stop with other components
878. 34:15okay so open flow this
879. 34:18approach was developed at the stanford
880. 34:20university
881. 34:21to manage traffic between routers such
882. 34:23as wireless access points and the
883. 34:25controller
884. 34:26so the open flow protocol is a basic
885. 34:29element in building the sdn solutions
886. 34:32okay so search for open flow and open
887. 34:35networking foundation
888. 34:37for more information on it now the open
889. 34:40stack
890. 34:41this approach is a virtualization and
891. 34:43orchestration platform
892. 34:45designed to build scalable cloud
893. 34:47environments
894. 34:48and provide an iaas
895. 34:53okay or eos solution openstack
896. 34:56is often used with cisco aci
897. 35:00orchestration and networking is the
898. 35:02process of automating and provisioning
899. 35:05of network components
900. 35:06such as servers storage switches
901. 35:10routers and applications so if you want
902. 35:13to learn more about openstack
903. 35:14go ahead and search it on the internet
904. 35:18now other components include interface
905. 35:21to the routing system
906. 35:23or i2rs transparent interconnection
907. 35:27of lots of links or trill you've got the
908. 35:30cisco fabric
909. 35:31path or fp and the ieee 802.1
910. 35:35aq shortest pass bridging or spb
911. 35:44okay so in traditional and sdn
912. 35:47architectures
913. 35:48okay so in traditional router or sewage
914. 35:51architecture the control plane and data
915. 35:52plane functions occur on the same device
916. 35:55so routing decisions and packet
917. 35:57forwarding are the responsibility of the
918. 36:00device operating system
919. 36:02so in sdn or software defined networking
920. 36:05management of the control plane is moved
921. 36:07to a centralized sdn controller
922. 36:10now the figure compares traditional
923. 36:14and the sdn architecture so on
924. 36:17traditional architecture if you will
925. 36:18observe
926. 36:19we have here three devices and each
927. 36:21device
928. 36:22combines the control plane and the data
929. 36:24plane now with the implementation of the
930. 36:26sdn architecture
931. 36:28this control plane from the traditional
932. 36:30architecture were moved out
933. 36:33okay and it is now
934. 36:36on the same location or centralized
935. 36:38location okay in this case the sdn
936. 36:40controller
937. 36:41which uses an open flow to manage this
938. 36:44devices here now the sdn controller is a
939. 36:48logical entity that enables network
940. 36:50administrators
941. 36:51to manage and dictate how the data plane
942. 36:54of switches and routers should handle
943. 36:55network traffic
944. 36:57it orchestrates or mediates and
945. 37:00facilitates communication
946. 37:02between application and the network
947. 37:06elements
948. 37:08so the complete sdn framework is shown
949. 37:10in the figure
950. 37:12so note the use of application
951. 37:14programming interfaces or apis
952. 37:16within the sdn framework okay
953. 37:19so an api is a set of standard requests
954. 37:23or standardized requests that define the
955. 37:26proper way for an application to request
956. 37:27services
957. 37:28from other application so the sdn
958. 37:31controller uses a northbound apis
959. 37:34but the northbound api is here okay
960. 37:38to communicate with the upstream
961. 37:39applications
962. 37:41this apis can help network
963. 37:43administrators
964. 37:45shape traffic and deploy services the
965. 37:47sdn controller also uses a southbound
966. 37:50apis so to define the behavior of the
967. 37:54data planes
968. 37:55on downstream switches and routers
969. 37:58so open flow is the original and widely
970. 38:01implemented
971. 38:02southbound api
972. 38:10so next is controller
973. 38:14right so sdn controllers and operations
974. 38:18so the previous topic covered sdn this
975. 38:20topic will explain
976. 38:22controllers now the sdn controllers
977. 38:25define the data flows between the
978. 38:27centralized control plane
979. 38:29and the data planes on individual
980. 38:31routers or switches
981. 38:33okay now each flow traveling through the
982. 38:36network
983. 38:36must first get permission from the sdn
984. 38:39controller
985. 38:40which verifies that the communication is
986. 38:42permissible according to the network
987. 38:44policy
988. 38:45so if the controller allows a flow
989. 38:48it computes a route for the flow to take
990. 38:52and add
991. 38:52an entry for that flow in each of the
992. 38:55switches along the path
993. 38:57so all complex functions are performed
994. 39:00by the controller so the controller
995. 39:03populates
996. 39:04flow tables so switch manage
997. 39:07the flow tables so in the figure the sdn
998. 39:10controller
999. 39:11communicates with open flow compatible
1000. 39:13switches
1001. 39:14using the open flow protocols
1002. 39:17okay so you've got an open flow switch
1003. 39:19here and it uses an open flow protocols
1004. 39:22to communicate among the switches
1005. 39:26now this protocol uses a tls or the
1006. 39:29transport layer security to securely
1007. 39:31send control plane
1008. 39:33communications over the network so each
1009. 39:35op and flow switch
1010. 39:37connects to other open flow switches
1011. 39:40they can also connect to the end user
1012. 39:41devices that are part of the patent flow
1013. 39:49now within each switches a series of
1014. 39:51tables implemented in hardware or
1015. 39:53firmware
1016. 39:54are used to manage the flow of packets
1017. 39:56through the switch
1018. 39:58so to the switch a flow is a sequence of
1019. 40:01packets that matches a specific entry
1020. 40:03in flow table now the three table types
1021. 40:07shown in the previous figure are as
1022. 40:09follows
1023. 40:10so you've got a flow table a group table
1024. 40:13and a meter table now the flow table
1025. 40:17this table matches incoming packets to a
1026. 40:19particular flow
1027. 40:21and specifies the functions that are to
1028. 40:23be performed in the packets
1029. 40:25so there are multiple flow tables that
1030. 40:28operate in the pipeline fashion
1031. 40:31group table a table may direct a flow
1032. 40:34to a group table so which may trigger
1033. 40:38a variety of actions that affect one or
1034. 40:41more flows
1035. 40:44mirror table this table triggers a
1036. 40:46variety of
1037. 40:47performance related actions on flow
1038. 40:50including the ability to rate
1039. 40:52limit the traffic
1040. 41:07okay so next is the cisco aci
1041. 41:11so very few organizations actually have
1042. 41:13the desire
1043. 41:14or skill to program the network using
1044. 41:16sdn tools
1045. 41:18however the majority of the
1046. 41:20organizations
1047. 41:21want to automate the network accelerate
1048. 41:23the application deployments
1049. 41:25and align their id infrastructure to
1050. 41:28better meet
1051. 41:29business requirements so cisco developed
1052. 41:31the
1053. 41:32application centric infrastructure or
1054. 41:34aci
1055. 41:35to meet these objectives in more
1056. 41:37advanced and innovative ways
1057. 41:39than earlier sdn approaches now cisco
1058. 41:42azi
1059. 41:43is a hardware solution for integrating
1060. 41:46cloud computing and data center
1061. 41:48management
1062. 41:49so at a high level the policy element of
1063. 41:52the network is removed from the data
1064. 41:54plane
1065. 41:55just simplifies the data center networks
1066. 41:57are created
1067. 42:02so there are three components or core
1068. 42:05components of the aci architecture
1069. 42:07which includes the application network
1070. 42:09profile or a and p
1071. 42:11an anp is a collection of endpoint
1072. 42:14groups or
1073. 42:14epg their connections and the policies
1074. 42:18that define those connections
1075. 42:20the epgs shown in the figure such as
1076. 42:23vlans
1077. 42:25okay web services and applications
1078. 42:28are just examples an anp
1079. 42:31is often much more complex so figure
1080. 42:34on the next slide now the next one is
1081. 42:37the application infrastructure
1082. 42:39controller
1083. 42:40or the apic now the apic
1084. 42:43is considered to be the brains of the
1085. 42:45aci architecture
1086. 42:47apic is a centralized software
1087. 42:49controller that manages and operates a
1088. 42:51scalable aci clustered fabric
1089. 42:53it is designed for programmability and
1090. 42:56centralized management it translates
1091. 42:58application policies into network
1092. 43:01programming
1093. 43:02okay so the last one is the cisco nexus
1094. 43:059000 series switches
1095. 43:07so this switches provides an application
1096. 43:10aware switching fabric
1097. 43:11and work with apic to manage the virtual
1098. 43:14and physical network infrastructure
1099. 43:19so the apic is positioned between the
1100. 43:22apn
1101. 43:23and the aci enabled devices
1102. 43:27okay so the apic translates the
1103. 43:29application requirements
1104. 43:30into network configuration to meet those
1105. 43:33needs
1106. 43:34okay within the network infrastructure
1107. 43:40okay so next is the spine lift topology
1108. 43:44the cisco aci fabric is composed of the
1109. 43:47apic
1110. 43:48and the cisco nexus 9000 series switches
1111. 43:50using a two tier
1112. 43:52spine lift topology as shown in diagram
1113. 43:54here
1114. 43:55okay now the leaf switches always attach
1115. 43:59to the spines
1116. 44:00but they never attach to each other so
1117. 44:03similarly
1118. 44:04the spine switches only attach to the
1119. 44:06leaf and
1120. 44:07core switches okay of course not shown
1121. 44:09here on the diagram
1122. 44:11now in this two tier topology everything
1123. 44:14is one hub
1124. 44:15from everything else okay so
1125. 44:19the cisco apic and all other devices in
1126. 44:22the network basically attached to lift
1127. 44:23switches so when compared to sdn
1128. 44:27the ap controller does not manipulate
1129. 44:30the data path
1130. 44:31directly instead the apic centralizes
1131. 44:34the policy definition
1132. 44:36and programs the leaf switches to
1133. 44:38forward topic
1134. 44:39based on the defined policies
1135. 44:45so the following are the sdn types okay
1136. 44:48so the cisco application policy
1137. 44:50infrastructure controller
1138. 44:52the enterprise module apic em extends
1139. 44:55the aci
1140. 44:56aimed at enterprise and campus
1141. 44:58deployments
1142. 44:59so to better understand the apec em it
1143. 45:02is helpful
1144. 45:03to take a broader look at the three
1145. 45:05types of sdn so the first one
1146. 45:07is the device based sdn
1147. 45:10now in this type of sdn the devices are
1148. 45:13programmable by applications running on
1149. 45:16in the device itself
1150. 45:17or on a server in the network as shown
1151. 45:20in the figure here
1152. 45:22okay so the cisco one pk
1153. 45:27is an example of a device based sdn
1154. 45:31it enables programmers to build
1155. 45:33applications
1156. 45:34using c language and java with
1157. 45:38python to integrate and interact with
1158. 45:41the cisco devices
1159. 45:45so the second type of controller or sdn
1160. 45:48type
1161. 45:49is a controller based sdn now this type
1162. 45:52of sdn uses a centralized controller
1163. 45:54that has knowledge
1164. 45:56of all the devices in this in this
1165. 45:58network
1166. 46:00okay now the applications can
1167. 46:03interface with the controller
1168. 46:06responsible for managing devices and
1169. 46:08manipulating traffic flows
1170. 46:10throughout the network so the cisco open
1171. 46:13sdn controller is a commercial
1172. 46:15distribution of
1173. 46:17open daylight okay
1174. 46:21so next is a policy based sdn
1175. 46:24now this type of sdn is similar to
1176. 46:26controller controller-based sdn
1177. 46:28where a centralized controller has a
1178. 46:30view of all the devices in the network
1179. 46:32as shown in the diagram here
1180. 46:34okay now policy-based sdn includes an
1181. 46:38additional policy layer that operates
1182. 46:40at a higher level of abstraction so it
1183. 46:43uses a built-in applications that
1184. 46:45automate advanced configuration tasks
1185. 46:48via a guided workflow and user-friendly
1186. 46:51gui
1187. 46:52so no programming skills are required
1188. 46:55the open
1189. 46:56or the apic em is an example of this
1190. 46:59type of
1191. 47:00sdn all right
1192. 47:06now this is an ap em features
1193. 47:10so each type of sdn has its own features
1194. 47:12and advantages
1195. 47:14the policy-based sdn is the most robust
1196. 47:17okay providing a simple mechanism to
1197. 47:20control and manage policies across the
1198. 47:21entire network
1199. 47:23now the cisco apic em is an example of
1200. 47:25policy-based sdn
1201. 47:27cisco em provides a single interface for
1202. 47:30network management
1203. 47:32including discovering and accessing
1204. 47:34device
1205. 47:35and host inventories viewing the
1206. 47:38topology
1207. 47:39as shown here on the diagram all right
1208. 47:42so tracing a path
1209. 47:43between endpoints and setting policies
1210. 47:51now the ap em pass trace
1211. 47:54allow the administrator to easily
1212. 47:56visualize traffic flows and discover
1213. 47:58any conflicting duplicate or shadowed
1214. 48:01acl entries
1215. 48:02so this tool examines specific acls on
1216. 48:05the path
1217. 48:06between two end nodes displaying any
1218. 48:08potential issues
1219. 48:10you can see where any acls along the
1220. 48:13path
1221. 48:14either permitted or denied your traffic
1222. 48:17okay so notice how branch router 2
1223. 48:21okay permits all traffic
1224. 48:25okay the network administrator can now
1225. 48:28make adjustments if necessary
1226. 48:30to better filter this traffic all right
1227. 48:34so we have reached the end of this video
1228. 48:37lecture
1229. 48:38have a great day see you on the next
1230. 48:44video