The Cloud Runs on Juniper | Juniper Global Summit

0:07 -Welcome to the cloud provider keynote session.

0:10 I'm Jason Fritch and I run sales

0:11 for Juniper's Global Cloud Provider segment.

0:14 I'm joined today by Mansour Karam,

0:16 Vice President of Products, and former co-founder and CEO of Apstra.

0:20 Welcome, Mansour.

0:21 -Thank you, Jason. It's good to be here.

0:22 -Juniper's been a market leader ever since the inception of cloud.

0:26 In fact, we are number one in cloud routing

0:29 according to leading cloud market research company,

0:32 the 650 Group.

0:33 In fact, in overall routing percentage of revenue

0:36 for the top five cloud providers,

0:38 Juniper leads all global competitors

0:40 and has been growing share for the last four years.

0:44 This position allows us to really understand your needs

0:48 and it allows us to see the future of networking

0:51 more clearly than our peers.

0:53 Over the last decade,

0:54 we've seen many cloud networking trends

0:57 emanate to the broader market.

0:59 You've been talking, we've been listening.

1:02 Here are some of the key cloud networking trends that we've noted.

1:05 First, massive scale and performance.

1:08 We're confident that we've got this covered

1:10 with routing platforms that reliably scale your network

1:13 and deliver cost-efficient, ultra high performance,

1:18 but we're not stopping there.

1:19 Second, disaggregation.

1:21 There's physical disaggregation, also known as scale-out.

1:25 This is favored in smaller,

1:26 fixed form factor platforms versus traditional chassis-based systems.

1:30 The advantages here include smaller blast radius

1:34 and the ability to add smaller increments of capacity.

1:37 There's also software disaggregation

1:39 which unbundles the networking OS from the hardware.

1:43 You can take this a step further by disaggregating the software itself.

1:47 We've done this with our new Junos Evolved operating system,

1:50 a cloud-optimized network OS

1:52 with microservices architecture that delivers resiliency,

1:55 feature velocity, and programmability.

1:58 It's important to note that the large cloud providers that we work with

2:01 have the institutional capabilities in automation and orchestration

2:05 to manage the increased architectural complexity

2:08 that comes with disaggregation.

2:11 Third, open, multi-vendor systems,

2:14 and a culture of experimentation that favors rapid execution

2:17 over long-term strategy and planning.

2:21 There's now in effect,

2:22 a series of cloud architectural principles

2:25 driven primarily by hyperscalers like Google, Amazon, and Microsoft

2:29 that go hand in hand with these cloud trends.

2:32 Mansour is going to talk a little bit more about these principles.

2:36 -Thanks, Jason.

2:38 Part of our founding mission at Apstra was to democratize these cloud principles

2:43 and develop some of these technologies

2:46 and package them as a turnkey product

2:49 so we could make them available to you,

2:51 every company that depends on the networking infrastructure

2:55 to run your business.

2:57 The first key principle that arises from these macro trends is abstraction.

3:03 There are multiple layers of abstraction.

3:06 There is a shift to an intent-based thinking,

3:09 outcome-based.

3:10 It's moving from imperative to declarative.

3:14 You tell the system the outcome you want to accomplish rather than how to do it.

3:19 Second, modularity.

3:22 This goes hand in hand

3:24 with the disaggregation trend that Jason mentioned.

3:27 Open APIs are key here,

3:29 as is the availability of software libraries for third-party developers.

3:34 This allows you to take advantage of the innovation of the larger ecosystem.

3:38 Third, telemetry.

3:41 Operations begins with the system data, the telemetry.

3:45 The goal is to instrument everything that needs to be instrumented

3:50 and consolidate into a single source of truth.

3:54 As more devices provide more telemetry

3:57 and as software-defined layers of infrastructure add management ease,

4:02 it becomes necessary to gather telemetry from the various infrastructure layers

4:07 and to capture the state of the network

4:10 and the relationships between the various layers.

4:13 The state is then used to provide unified visibility,

4:17 to determine the root cause of anomalies, and to quickly troubleshoot problems.

4:23 -Thanks, Mansour,

4:24 and if you're operating your data center,

4:26 these capabilities are critical to keeping up with business demands.

4:30 Now, as you know, not everything is moving to public clouds.

4:33 The world is and will continue to be multi-cloud and hybrid cloud,

4:37 consisting of multiple public clouds as well as on-premises data centers,

4:42 depending on the workload.

4:44 This is why bringing these cloud principles to all companies

4:47 that operate data centers is so important.

4:50 Today, success in the cloud is more important than ever

4:52 because application software

4:54 is the determining factor in technology user experience,

4:58 which is at the center of staff and customer experience.

5:01 Most of that application software runs on data center infrastructure,

5:04 whether it's on-premise or in a public cloud.

5:07 It's not hard to understand

5:08 that the challenge of building and operating data centers today

5:11 is less about total cost of ownership and more about revenue and risk.

5:16 I began here today by telling you that

5:18 we've been the dominant player in cloud routing for years.

5:22 As you've heard Rami earlier and David Cheriton,

5:25 we're doubling down on simplifying data center operations as well.

5:28 Mansour, why don't you now take us through how we're doing this.

5:33 -Yes, absolutely, Jason.

5:34 You're right, not every workload will be run on the public cloud.

5:39 In fact, it is often far cheaper,

5:43 more secure, and more reliable to run a private cloud.

5:47 This is true as long as the network infrastructure is automated effectively

5:53 to provide a simple and predictable operational model.

5:58 This makes your networking automation strategy crucially important.

6:03 You need the data center infrastructure

6:05 that will enable you to run at the speed of the business

6:10 and automation software is key to make that happen.

6:14 If you're a do-it-yourselfer,

6:16 you could use Apstra to deliver a turnkey automation capability,

6:21 which you can then integrate through Apstra's open APIs

6:25 with your CI/CD pipeline and the other components in your infrastructure.

6:30 This is what Yahoo Japan has done,

6:33 utilizing the Apstra solution

6:35 to fully automate and operate their network while leveraging Ansible

6:40 to integrate Apstra with the rest of their workflows.

6:44 Yahoo Japan also uses Apstra's streaming telemetry

6:48 to stream the telemetry that Apstra has collected

6:52 through third-party databases such as InfluxDB.

6:57 If on the other hand, you want automation software

7:00 that will fully automate your network turnkey,

7:04 then the Apstra solution is ideal for you.

7:07 In fact, we have customers that have gone from zero

7:11 to a fully operational automated network in 30 minutes.

7:17 Regardless of the type of customer that you are,

7:20 we found that everyone needs software that allows small teams

7:25 to efficiently operate and scale their data centers.

7:30 In either case, you'll benefit from the power of intent-based networking

7:35 and its simple and predictable operational model.

7:39 Our customers have realized 83% reduction in OPEX,

7:44 99% improvement in agility,

7:48 and 70% reduction in mean time to resolution with Apstra.

7:54 If you're struggling with agility versus reliability

7:58 or concerned with vendor lock-in,

8:01 we can help.

8:03 Our mission is to deliver the hyperscaler operational experience

8:08 to all cloud companies,

8:09 all enterprises, and all network service providers without vendor lock-in.

8:16 Our mantra is, Experience-First Networking

8:20 for data center operators, you, and their end-users.

8:26 This is built on a foundation of intent-driven,

8:30 closed-loop automation, and remediation for the entire service life cycle

8:36 with a unified toolset across architects and operators, starting with day zero, design.

8:46 With telemetry at the base to automate capacity,

8:50 forecasting, and demand planning,

8:52 and pre-validated the designs that install quickly,

8:56 we help you with a disciplined approach

8:59 to designing data centers in a repeatable fashion.

9:03 Then comes Day 1, deployment.

9:07 Abstraction is critical here.

9:10 Apstra is the pioneer in intent-based networking.

9:14 When you abstract out the infrastructure

9:16 so that the services are driven by blueprints,

9:19 your solution is inherently multi-vendor.

9:23 Data center operators describe the "what"

9:26 and our blueprints deliver the "how",

9:30 pushing configurations to the infrastructure

9:32 while validating that the network is indeed working as intended.

9:38 Complexity is the enemy of improving all factors of cost, revenue, and risk.

9:45 Simplicity is the cure and it is at the core of everything we do at Juniper

9:50 to remove abstract out and automate out complexity.

9:55 Simplicity is also a prerequisite for reliability and predictability.

10:03 For example, Beelastic, another Apstra customer,

10:06 went from a proof of concept to full production in two weeks.

10:13 Then, last but certainly not least, Day 2.

10:18 The ongoing operations and management of the data center fabric.

10:23 You have moves, adds, changes.

10:27 You need to troubleshoot issues when they arise.

10:30 You need to do software updates.

10:33 EVPN-VXLAN data centers are not easy to manage or operate.

10:39 Ansible templates proved useful and they solve some problems, but not enough.

10:45 We've proven that Apstra reduces the time and personnel

10:48 required to manage your data center network.

10:51 The blueprint is the single source of truth,

10:54 so the system can continuously validate operations to match this.

11:00 Apstra continuously compares the operational state with the intent.

11:05 The delivered servers must match the intended blueprint.

11:10 This is closed-loop assurance,

11:13 and Apstra provide guardrails when changes are required.

11:18 Apstra also pinpoints the root cause of problems and instantly isolates failures.

11:26 We lower the meantime to innocence

11:28 when end users are complaining about downtime.

11:31 Apstra decouples the management plane from the infrastructure,

11:35 so it's inherently open and multivendor.

11:39 Set your operational model around what you need, not a vendor.

11:45 The end result is scalable and simple to operate a EVPN-VXLAN networks

11:50 for a wide range of environments and independent of size.

11:55 Be sure to watch the Apstra demo in the data center track

11:59 and you can see all of this in action.

12:02 What this also means is that you get to do more with far less

12:08 and dramatically reduce the probability of making costly mistakes.

12:13 This means peace of mind

12:15 and more time focusing on other areas that matter to you.

12:20 If you're like me, this means spending more time with your family and kids.

12:24 If you missed out on dinner with family

12:27 because you were troubleshooting some problem,

12:30 or if you had to stay late at night to manage a network update,

12:35 then this is not too good to be true.

12:38 We can help.

12:40 How does this all sound to you, Jason?

12:43 -It sounds great, Mansour.

12:45 What it means to you is simple.

12:47 A smooth data center and network operations experience,

12:50 and ultimately, faster innovation.

12:53 Please be sure to watch our customer interviews,

12:55 the data center content track, as well as the rest of the show.

12:58 Thanks for watching.

12:59 -Thanks for watching.