Machine Learning: Google’s Route to their Enterprise Cloud Ambition

Google Cloud Platform (GCP) is growing quite rapidly, quarter by quarter, said its SEA head, Tim Synan. “By some accounts, we are the fastest growing cloud (service), quarter-on-quarter, and we are delighted about that.”

Google’s cloud platform product has been around since 2011, but it was only when VMware founder, Diane Greene became head of Google Cloud in late 2015, did the young company start to show its potential.

Since then, USD30 billion has gone into beefing up the platform’s already expansive infrastructure, and very big brand names already count as some of their customers –  HSBC, Colgate, Verizon, eBay, Snapchat – with many more to come in the future.

Synan noted, that in Southeast Asia, there is some 135-percent increase in paying cloud customers year-on-year, a fact that is backed by Google having launched its first region in SEA last year, via three zones which are all located in Singapore.

This business of building cloud regions will be replicated to five other locations as well – Netherlands, Finland, Hong Kong, Los Angeles, Montreal – with the objective of improving round-trip-time latency, for customers located in Google’s regions. This was announced only 3 months ago.

For Southeast Asia, Synan noted that GCP continues to see a broad mix of companies working with them, on a diverse set of workloads, from productivity and collaboration solutions via G-Suite, all the way to fintech payments solutions.

“A lot of that has to do with people’s comfort level with cloud technologies. It’s certainly much more reliable now, and (there’s the) innovation that people can tap into to solve really hard problems around data, and machine learning in particular; it’s really hard for organisations to build those capabilities up on their own.”

Synan also believes that Google has improved the security posture of a lot of organisations. “They work with us because they can achieve better security, and particularly with Google Cloud, we have a competitive advantage around security.

“That’s been a driver for some workloads as well.”

Machine learning and advanced analytics

Cloud is also proving to be an effective way to democratise certain technologies like machine learning.

Synan pointed out, “Machine learning requires a lot of compute, a well-trained model and a lot of data. And that’s why the cloud is great for machine learning.” Google’s TensorFlow helps move things along, enabling organisations to, very simply put it, build deep learning models.

Google has also started a resource hub called Learn with Google AI, which has a collection of education resources ranging from core machine learning concepts, to developing and honing machine learning skills, and then applying them to real-world problems.

They work with us because they can achieve better security, and particularly with Google Cloud, we have a competitive advantage around security.

Google’s blog described this resource as, “From deep learning experts looking for advanced tutorials and materials on TensorFlow, to “curious cats” who want to take their first steps with AI, anyone looking for educational content from ML experts at Google can find it here.”

It also features a new, free course, “Machine Learning Crash Course (MLCC).”

In March last year, Google acquired Kaggle, a crowdsourcing platform for predictive modelling and analytics that use public-hosted datasets. Kaggle was described as a platform for data science on-demand, which spans nearly 200 countries.

“One of the things we want to do here at Google, is democratise how people are using AI,” Synan explained.

“This is our approach to machine learning – our cognisance that not everyone has data scientists and real advanced analytics capabilities within their organisations.”

So, on one spectrum there is pre-trained models leveraging data from Google, that are made available to organisations and developers who can take those pre-trained models and embed them into their applications today, via APIs.

On the other extreme, are organisations who want to build their own models, and one option to do this is by partnering with Google. “That’s where we have TensorFlow (to address this).”

This is our approach to machine learning – our cognisance that not everyone has data scientists and real advanced analytics capabilities within their organisations.

Google recognises also the goldmine of data that they have, and the marketing analytics opportunity it brings.

“That’s something we work quite closely with our colleagues on Google Cloud – around understanding customer behaviour, being able to segment that … and we do all that by taking all the rich analytics that other Google products generate, and manage and analyse that under GCP,” Synan said.

Unique advantages

Synan said that Google has the advantage of benefiting from an already mature market, but they are fundamentally also doing some things differently.

“One of it, is the way we partner.

“We like to talk about our engineer-to-engineer capability, and when we work with some of our large customers, they appreciate the way that we can rally and assemble real technical skills that matter to their workloads.

“So, it’s an engineer-based partnership, and we continue to focus upon that.”

Other advantages Synan believes their customers benefit from is Google’s openness. “We are real believers and we are trying that agenda across cloud. Kubernetes, TensorFlow, Apache Beam and so on, we will continue to ensure that developers have choice.” The last six months alone, Google Cloud has also shipped 500 releases.

There is also all the investments into submarine cable networks that is paying off. “The Google Cloud network is a tremendous asset we have and organisations are starting to realise that.

“When you leave the Google Cloud region, you are on the Google Cloud network until the very last point-of-presence closest to the end user. There are security and performance and quality of service benefit here,” Synan said.

Google also boasts a commercial model which doesn’t require signing of contracts, making it easy and frictionless for large companies all the way to individual developers, to use Google Cloud.

Late to the game?

Someone rightly pointed out that the cloud computing industry, especially at the global hyperscale segment, can do with more players.

Amazon Web Services or AWS, is the player that every other cloud player is taking aim at, and Google certainly has the heft and the clout to participate in an industry with a very prohibitive barrier to entry. Not many other tech companies could match AWS’s infrastructure expansion, data centre-wise. On top of that Google has a first mover advantage among the other cloud players, as a submarine cable owner.

It started with the Trans-Pacific Unity cable in 2009 and it hasn’t stopped there. Google announced in January, investments into three more cables, making the total number of cables they directly invest in to date, eleven. With these expansive network infrastructure, up to 40-percent of the world’s Internet traffic, can be traversing Google’s proprietary networks.

At this point, it’s clear Google will do whatever it takes to climb the ranks and overtake Amazon Web Services.

But there is one more cloud player that should not be overlooked.

The might of the Chinese commerce player, and now also cloud player with Alibaba Cloud, cannot be underestimated.

In Malaysia specifically, they have already launched an artificial intelligence product that is being used by the public sector, in a smart traffic/smart traffic scenario. With one foot firmly in the door, this project aims to scale to other cities, effectively also serving as a nation scale proof-of-concept that Alibaba can leverage as they bring it the smart cities vision to other countries.

Notably, Alibaba also has a platform called Tianchi, which is similar to Google’s Kaggle and which calls itself a global big data crowd intelligence platform that holds contests.

The might of the Chinese commerce player, and now also cloud player with Alibaba Cloud, cannot be underestimated.

Also, Google’s pre-2013 plan to build a data centre in Hong Kong which had been canned, was revived once again this year.  Building a cloud data centre that close to China, seems to indicate Google wants to take the fight to Alibaba Cloud, at least in the area of artificial intelligence.

But, besides being known for being extremely good at AI and democratising the use of it for more and more businesses to leverage, what is Google Cloud going to shape up to be?

The cloud industry

Google’s relative late entry into the cloud game, was exacerbated by their first few years without good leadership at the helm.

To date, AWS is the cloud player to beat; it has a larger tool set for developers, a bigger ecosystem of partners, and enterprises that have gone all-in with AWS.

Will we see Google Cloud making as huge an impact upon the enterprise landscape; and a legacy one at that; as what AWS is doing with the database industry?

Even Oracle, themselves a substantial cloud player, sees AWS as a big threat, because Oracle’s on-premise customers are finding Amazon’s cloud to be a viable home for Oracle databases.

Moving an Oracle database to AWS is the first step to dumping Oracle completely; some observers have opined; and finally loosen Oracle’s grip upon database customers.

Will we see Google Cloud making as huge an impact upon the enterprise landscape; and a legacy one at that; as what AWS is doing with the database industry?

It’s too early to tell.

But during EITN’s interview with Synan, he voices Google’s commitment to continue to invest in and expand their infrastructure footprint.

“Data, machine learning, Internet of Things, our infrastructure, are still important to us, and we will continue to work on those,” the SEA head concluded.