Building a Business Case for WAN Automation
The pros and cons of DIY network automation – is it worth it?
For the first time, service providers are showing positive results around customer satisfaction and network automation is to blame. Dr. Ray Mota of ACG Research explains why a change of mindset is required to increase automation for business continuity and service profitability.
"Do you have automation or silos of automation,” Dr. Mota asks before sharing new key statistics and insights into taking automation beyond task management and into the realms of operational efficiency and business agility.
Watch to learn the pros and cons of these different network automation approaches: packaged solutions, customized solutions, or DIY.
Who is paying 1 cent vs. $600 for annual server management
Why your competitors are really your partners
Two questions board members always ask Dr. Mota about $$$
Who is this for?
0:08 Hi everybody. My name is Brendan Gibbs.
0:10 I'm Vice President of Juniper's Automated WAN Solutions business.
0:13 I'm joined today by Ray Mota,
0:15 CEO & Principal Analyst of ACG Research.
0:17 Ray, thank you so much for joining us.
0:19 Thank you for the invite.
0:20 Appreciate it. The weather's perfect over here in California.
0:22 It’s called Sunnyvale for a reason.
0:24 That's a good point.
0:25 Ray, I always love talking to you because you have such a unique perspective
0:28 on the service provider industry overall.
0:30 You talk to everybody.
0:32 One thing I just would love to learn from you is
0:34 what are you seeing in terms of service provider uptake of automation?
0:37 What are they doing?
0:38 It’s a good question.
0:40 I think it's important to understand the rationale with it.
0:43 I would say, over the last two years, has become a high priority.
0:47 Right now,
0:47 they're looking for simplification and automation across different parts.
0:52 Not just in network automation,
0:54 but business automation and upending the process.
0:58 A lot of that is driven because their customers, for the first time,
1:02 are accelerating the digital transformation.
1:05 It used to be that I used to talk to a lot of this enterprise
1:08 about digital transformation and it was more slideware.
1:12 Now, they're accelerating because business continuity and service profitability.
1:17 It was good to see that service providers are reacting to that piece
1:21 because they have a risk at that pace.
1:23 They could miss the inflection points for that area.
1:26 For the first time, we're seeing not only them implement it,
1:29 but seeing positive results.
1:30 We recently did a survey where we saw over 72% actually saw an increase
1:36 in customer satisfaction of the service that they're delivering
1:40 because of network automation.
1:42 The time to deliver new services is over 52%.
1:46 Then the OPEX which was the biggest concern is actually,
1:50 on average, between 32% to 40% increase,
1:53 when you start looking at automation more than just task management automation.
1:58 When it gets to related to operational efficiency,
2:01 operational agility, and business agility.
2:04 Wow, that's an amazing set of statistics.
2:06 Clearly, it's having a big impact.
2:08 I wonder, you mentioned, in the last two years, it's accelerated,
2:12 do you think any of this is driven by COVID as in the SPs
2:15 found themselves just stuck at home all of a sudden?
2:18 Yes, I think that's a big driver for it,
2:20 because with this new hybrid work at home strategy,
2:24 they realized, I can't get to the office like I could.
2:28 Even things like segment routing,
2:30 simplifying networks, things like different parts of automation.
2:34 I think the biggest thing that a lot of them realized is there were a few,
2:37 Brendan, that actually thought they had automation,
2:40 but they had silos of automation.
2:42 That became difficult to manage remotely.
2:45 A lot of them realized that they needed to look at cross-domain
2:48 and change their mindset to start thinking about the service that I deliver.
2:52 What are the services and how do I automate from an end to end?
2:56 Like security.
2:57 Security is as strong as your weakest link.
2:59 When you look at a security domain,
3:01 you look from an end-to-end type aspects of it.
3:04 Automation should be not just physical assets,
3:07 it should also be some of the virtual assets together,
3:10 especially as you start moving to this hybrid cloud SaaS, physical environment.
3:15 I think, when you start removing those silos of automation,
3:18 that's where you're seeing the positive impact there.
3:20 Wow, that's amazing insight.
3:22 You also mentioned enterprises.
3:25 Are enterprise customers of SP,
3:27 do you think are they driving the use case for automation
3:31 or are there other use cases they're really responding to?
3:34 No, there's some service provider use cases itself.
3:36 There's the part of the disaggregation that's going on in mobile?
3:41 Open RAN, vRAN is becoming a high priority,
3:44 and those are great from a cost CapEx,
3:46 but if you don't bring automation to address the operational efficiencies
3:51 in there and the risk exposure,
3:54 when you do things like network slicing that allows you to scale and offer new revenue,
3:58 you're going to create an operational nightmare.
4:01 Service providers need to understand,
4:03 as they go to deliver these services,
4:05 Private G is another use case and it's a high,
4:09 which is more vertical enterprise.
4:11 Network automation needs to be a thought process from day one.
4:15 I use the analogy back-- I can probably mention.
4:18 If you look at a server AT&T versus a server at Google,
4:22 to manage one server at AT&T, just management,
4:26 it was about $600 annually.
4:27 It doesn't sound like much.
4:29 When you look at the number of servers they had, that adds up to a pretty number.
4:33 That same server at Google would be $0.01.
4:37 $0.01 versus $600, wow.
4:39 Because everything from day one is automation,
4:42 simplification, visibility, and control.
4:46 Wow, that's a huge disparity.
4:49 That just is mind-blowing.
4:50 It needs to be solved.
4:52 They need to solve it.
4:53 How does service providers think about this then?
4:55 They look to see what's worked.
4:57 I think, if you look at some of what the web scalers have done,
5:00 they've done a good job to say,
5:02 "You need to start thinking of automation from day one.
5:06 You need to be able to have a hybrid environment across."
5:10 It's good for service providers because they have a lot of legacy,
5:13 but now because there's these transitions, and investments in 5G,
5:18 they realize that consumer 5G, the business case isn't quite there yet.
5:24 They can't justify the CapEx spending.
5:26 As they look for ways for, let's say,
5:28 Open RAN to work on the CapEx reduction,
5:31 they really need to look to offset the investment by looking at things like private 5G,
5:37 but look at what web scalers is have done in the past
5:39 with the plug and play and end to end mindset.
5:42 They can literally learn a lot.
5:44 A lot of them now are looking at,
5:46 when you look at what you call Industry 4.0,
5:49 where you look at certain verticals,
5:50 that's something that there's a part in there that web scalers play well,
5:53 which is a telco cloud there, capacity and compute.
5:57 The service writers primarily focus on the connectivity layer,
6:01 which is 5G, or wireline.
6:03 There's other layers on top of there,
6:05 edge services layer and then you have applications layer.
6:09 Web scalers have been good at creating this ecosystem environment.
6:13 There's a lot of lessons and a lot of these service providers need to figure out,
6:17 how do I become web light-type customer there?
6:23 I remember thinking, a few years ago,
6:24 that service providers and public cloud operators
6:27 or hyperscalers were like oil and water,
6:30 especially as you say, telco cloud, a lot of operators, it seems to be,
6:35 without being too dramatic, they saw the public hyperscalers as competitors.
6:41 Of late, we've seen a number of partnerships.
6:44 Now, I'm wondering, given what you're saying,
6:46 that massive disparity and also a little bit more closeness between them,
6:50 is there a role for public cloud in an SP automation world?
6:54 Yes, there is, but they need to be careful.
6:55 There's a mix. You're going to see a mix.
6:57 I think, from a service provider point of view,
6:59 you need to make sure, "Do I have the skill sets?"
7:01 Number one.
7:02 Number two, "Who wants to own the customer?"
7:06 This is an important part that they need to be careful with.
7:09 You're seeing AWS, Google, Azure,
7:12 go in there and on the edge services to say,
7:15 "We can get your cloud-ready by offering that piece."
7:19 There's a revenue share opportunity that they can do in that area.
7:22 Service providers need to be realistic also to say,
7:25 "When I do revenue share,
7:26 who's going to be responsible for the customer?"
7:29 The service provider has the skill sets.
7:31 The challenge, I think,
7:32 has been is that these service providers need to find some solutions to say,
7:36 "Could I implement network automation myself to do DIY?"
7:41 If they have the skill set or find the right partner to help them with that,
7:45 then they got a solution to say,
7:47 "In this particular service, I can partner.
7:49 This one, I want to do DIY."
7:51 Finding a vendor that could help them with that network automation
7:54 in turn becomes an important factor.
7:56 Well, do you think that service providers are more open to vendors,
8:00 whether they be a big public cloud operator or a vendor,
8:03 like a router vendor, or a switch vendor, or optical vendor,
8:06 offering some package solution, or are they all still thinking,
8:10 "I need to build my own automation suite from scratch"?
8:13 It’s a good question.
8:15 I think, when you look at that part of it, it's a mix. honestly, it's a mix.
8:19 I've seen some work successful and I've seen some nightmares,
8:23 because what a lot of the service providers realize is that
8:26 when they build their own product,
8:28 someone's got to support it, someone's got to maintain it,
8:31 someone's got to bring the staff with the right skill sets
8:34 and be able to keep that staff happy to do that.
8:38 In a way, they have to become a vendor,
8:40 like who do they call when a problem happens?
8:41 They look at themselves.
8:43 Whose neck do they choke?
8:45 They choke themselves.
8:46 I think a lot of service providers that have gone in that route go,
8:49 "Timeout, I don't want to choke myself.
8:52 I need to find somebody else from that."
8:54 I always talk about where I was the naive CTO,
8:57 wherein Linux first came out and it was one of those things we're like,
9:00 "Well, I could save my company millions of dollars.
9:02 I don't need to go with this particular operating system."
9:06 Then I realized, what happens when something breaks when I do DIY?
9:10 It would have been my neck to choke.
9:12 Then I found a vendor's neck to choke but they also educated me on more pieces
9:16 where I wasn't an expert like the optical layer and other pieces for that area.
9:20 I say it's missed because there is a handful that do have the right skill sets but I think,
9:25 from a longer or bigger term perspective, is better to partner and not become a vendor.
9:31 Do what you do good, and focus on that,
9:34 and find partners that you trust that could deliver
9:36 and be there for you in case of some support issues associated with that.
9:40 I think it makes a ton of sense and just taking it together,
9:42 a few things that you've said already,
9:44 it almost seems like
9:45 there's different layers to think about from a DIY perspective, do it yourself.
9:49 There's the telco cloud layer, maybe that doesn't need to be done as often.
9:52 As you were just saying,
9:53 maybe the software layer on top of that doesn't need to be customized.
9:56 It's more the services,
9:58 the unique aspects that will be customized for an SP, right?
10:01 Yes, that's correct.
10:02 I think, what a lot of service providers,
10:04 the one that have been successful in realizing
10:06 that the service that they're considering delivered
10:09 needs to be agnostic of the infrastructure.
10:13 If that's correct, then that's where you bring in automation from
10:17 an end-to-end point of view.
10:19 Any service provider that could get to that level
10:22 is going to be able to deliver services much quicker.
10:24 When you have the first saying,
10:26 "Well, this is the architecture I've got.
10:28 These are the only services I deliver," then all of a sudden,
10:30 your timeframe to deliver the services becomes a problem.
10:34 Actually, I want to key on that exact point in terms of timeframe.
10:38 Time to service, quality of deployment, et cetera,
10:41 is now being top of mind for everybody.
10:43 I'm very interested in your thought.
10:45 I wonder if there's an opportunity for service providers
10:47 to have a mind towards that time to deployment, time to service reality,
10:52 and start consuming things as a service from other core like SaaS, for example.
10:56 Some of it depends on certain factors based on the cutbacks they have.
11:01 Then the timeframe and miss opportunity,
11:03 because customer acquisition costs,
11:06 the longer you wait to deliver the service,
11:08 the longer it goes up customer acquisition costing.
11:11 That's a number one priority for board-level members and service providers.
11:16 Then the lifetime value of the customer is reduced if you miss it.
11:20 What ends up happening, let's say,
11:21 it's a managed service contract, if you're late to market,
11:24 you either have to buy out the contract,
11:27 that increases the costs, or you have to wait till that contract expires,
11:32 and then your delivery of time.
11:34 It's really important to have a situation
11:36 where you could plug and play and get to market quicker,
11:39 especially at the pace that your enterprise customer
11:42 because they'll move without you, at this point.
11:44 They move without you.
11:45 It makes sense.
11:46 What are some of the key factors for SPs?
11:49 For example, you mentioned a few key attributes in terms of improving time to market,
11:55 time to service delivery, reducing costs, maybe reducing complexity.
12:00 The answer may be all of the above, but is there a primacy there?
12:04 How would SPs look at this?
12:05 They want to optimize around one of those factors, the others are nice to have.
12:09 I think there's two discussions.
12:11 Anytime I go into a board meeting for the service providers,
12:15 if I don't have one of the two, it's not going to be a good meeting.
12:19 The first one is how do you help me save money?
12:22 What solutions or group of solutions could you help me to optimize my profitability?
12:28 That's the first,
12:28 so Open RAN or vRAN would be a discussion associated with that.
12:32 Then the other part is, how do you make me money?
12:35 How do you increase my TAM?
12:37 That's where I identified things like private wireless or private G
12:41 to say these are some low-hanging fruits, you have the infrastructure,
12:45 right now you need to expand that into an area and develop a go-to-market.
12:48 Those are the two areas.
12:50 How do you either make me new money or increase my TAM,
12:53 or how do you reduce my OPEX?
12:56 Network automation applies to both of those there.
13:00 I'm wondering also then,
13:01 how does, say, the quality of experience play into the latter case, that,
13:07 how do you make money?
13:08 Because I'm just thinking about how do SPs use automation not just to reduce their costs,
13:12 but also as a competitive differentiator, rather?
13:15 How do you get quicker time to market?
13:16 How do you deliver better quality experience to your customers?
13:19 How do you basically use that as a competitive weapon?
13:22 I think, quality experience is something,
13:24 me being a former CTO, that I was really naive about.
13:27 We recently did a survey and it was a number two priority,
13:32 customer experience, when it got to prioritization.
13:36 Customer acquisition cost was number one.
13:39 The other one is, how do we increase and maintain?
13:41 Because they spend so much money to acquire the customer.
13:46 A big priority is how do I maintain that customer?
13:49 The reason they lose it, is because of customer experience.
13:53 There are enough options out there with systems integrators,
13:56 and others to say, "If you're not going to give me customer experience,
13:59 I'll either find somebody else because things could run in the cloud
14:03 or I'll find an integrator and I try to do it myself there."
14:06 That's been the number two.
14:08 It was a eye-awakening for myself, but it is an important factor.
14:12 Yes, I actually wouldn't have dreamed it was a surprise.
14:15 No same, here.
14:16 It's compelling.
14:17 What comes next then?
14:18 You've already said that one of the key learnings
14:20 has been you need an end-to-end approach to break down the silos,
14:23 I think that was what COVID taught, what happens next?
14:27 I think when you start looking,
14:28 I think, from a future outlook perspective, there's a few things.
14:32 If you look to say,
14:32 "Okay, now I'm looking at things from a service point of view
14:35 and an end-to-end point of view."
14:37 If you look at things like machine learning,
14:39 AI, me and you are old enough to understand those protocols
14:43 and algorithms have been around for 20 to 30 years in some cases,
14:47 but the problem is you needed a Ph.D.
14:50 in-house to figure out how to implement it and get it to work
14:54 and see the value and that used to take years.
14:56 Now, these things run in the cloud.
15:00 Because they run in a cloud,
15:02 you could look at it from a cross-domain perspective
15:05 and you could work on what you call the input parameters, what do I need?
15:08 You send it to the cloud, does the complexity for you,
15:12 and you're able to do that more efficiently.
15:14 I think that's the next part of saying,
15:17 "This is where the future's going."
15:18 Moving a lot of that intelligence to the cloud,
15:21 and then I could look at things from a cross-domain and use it when I need it,
15:24 and stuff like that, so a lot of innovation going on in that space.
15:28 I love that.
15:29 I just have one last question then.
15:30 Just keying off what you were just saying,
15:32 it almost seems like, and this is a question then that,
15:36 it seems like what you're positing is using automation to take highly complex topics,
15:41 but make them easily consumable by the masses,
15:44 so to speak, as you say,
15:46 AI and ML and other complex algorithms,
15:49 you shouldn't need a Ph.D. on your staff to use automation,
15:52 so it almost speaks more to automation as an easy button
15:56 rather than say automation as a software developer,
16:00 DevOps or Ph.D. level old complexity, is that true?
16:03 Yes, that is correct.
16:04 I think you're seeing areas of automation and AI machine learning.
16:12 Like I said, I'm old enough to remember people used to talk about it,
16:14 but it was just too complex.
16:16 We're seeing, in some cases,
16:18 13-year-olds develop iPhone app where they would put in the input parameters,
16:23 send it to the cloud and the cloud does all the hard work whether is Z analysis.
16:27 Whatever type of algorithm they're using in there,
16:30 they don't care and they don't want to know,
16:32 just like that service provider is developing a service
16:35 and the underlying infrastructure should say,
16:38 "Plug whatever service you want."
16:40 We've become a platform for innovation,
16:42 and because we have automation that includes network automation,
16:47 security as well for all the provision and aspect,
16:50 we handle all the complexity for you.
16:52 You focus on, how do you deliver the service?
16:55 How to develop your go-to make?
16:56 How to price it?
16:57 Those type of things there.
16:59 That's super compelling.
17:00 This has been a great discussion, Ray.
17:01 My takeaways are automation is paramount and is mandatory,
17:05 simplicity is going to be the name of the game as well as end-to-end without any silos,
17:09 incorporating security, incorporating that kind of holistic focus
17:13 but becoming a springboard for future innovations as you're saying.
17:17 It's exciting times.
17:18 Absolutely. Absolutely.
17:19 Thank you for the invite there.
17:20 Thank you, Ray. This has been great.
17:21 Thanks to everyone for listening as well.