13 Mar Google Cloud Next ’17: Impressive, But Hard To Find Definitive Differentiators After Day 1
I attended Google’s annual, enterprise-focused event, Google Cloud Next ’17, last week in San Francisco, paying close attention to all the announcements coming out of the three-day event. I attended their industry analyst event as well, held the day before. The conference is Google’s chance to showcase its achievements from the past year, as well as announce new products and services within their cloud and collaboration portfolio. Competition for the public cloud and collaboration is intense with giants like Microsoft, Amazon.com AWS, Cisco Systems and IBM SoftLayer looking for their piece of it. Here’s my take on some of the key items coming out of Day 1 of the conference—it’s a lot, so let’s dive in.
Machine Learning announcements
Google started off by explaining their desire to “democratize AI.” I think that’s a great goal, but I wouldn’t have used the same exact words as Microsoft’s Satya Nadella used at Build, as it makes Google appear like a follower, which they aren’t. Google is one of the leaders in machine learning and they should have used a different phrase.
There were several updates in the machine learning realm. Google announced their Cloud Video Intelligence API, which allows developers to search for specific content within videos with noun or verb search terms—like “cat” or “swim.” The API utilizes frameworks (TensorFlow for example) to apply deep-learning models to video-streaming media platforms. I think this is honestly pretty cool—there’s a lot of unstructured, unutilized data floating around out there in video, and this API will allow developers to mine it in a way they never have before. This project is currently in the private beta stage. It’s obvious Google took many of their learnings from their indexing work evident with YouTube and offer it to businesses. I see Microsoft and Google currently leading in the self-serve ML API space.
Next, Google announced their Cloud Machine Learning Engine, which it said will help customers train and deploy their own machine learning models within the cloud. It has the potential to function like a managed service for constructing TensorFlow-based, scalable models capable of interacting with any type of data. They’ve integrated it with their Cloud Platform’s data analytics portfolio: Cloud Dataflow (used for data processing), Cloud Datalab (used for data science workflow), and Google BigQuery (which performs SQL analytics). They’ve also strategically partnered with other tech companies like SpringML and SparkCognition to power their own custom analytics solutions. I see this space as a three-horse race between Amazon.com AWS, Google and Microsoft and maybe even IBM with its POWER solutions.
Google also announced an update to their Cloud Vision API, Version 1.1, used to extract metadata from images. This API has been extremely popular, and is growing at a very quick rate—this update will help round out the API, allowing customers to classify a more diverse set of images. They also introduced their Cloud Jobs API, a deep-learning model to power job-search websites with more relevant results. For example, it allows users to search jobs based on commute time and transportation mode—a feature so handy, I’m kind of kind of amazed it didn’t exist until now. To fine-tune the API, Google has opened it up to testing by job-search services (such as Jibe, Dice, and CareerBuilders), and utilized their feedback in the design. Cloud Datalab, mentioned earlier, also received a GA release—improving scientific workflow by allowing the analyzation and visualization of data within CloudStorage, local storage, or Big Query. It comes TensorFlow and Scikit-learn ready, and for batch and stream processing, it is supported by Cloud Dataflow and Apache Spark via Cloud Dataproc.
Another big announcement coming from Day 1 was a new strategic partnership with business software powerhouse SAP, aimed at developing enterprise solutions through the integration of Google’s cloud and deep learning technologies with SAP’s enterprise applications. SAP has partnered with nearly every other public cloud service and the industry was all awaiting this announcement. It also represents SAP’s march away from their IaaS to SaaS and PaaS layers and finding partners for AI.
The partnership will be multifaceted—it includes certification of SAP’s in-memory database SAP HANA on the Google Cloud Platform, integrations with G Suite, data governance collaboration, and machine learning collaboration. The tie-up may not be unique, but I see this partnership as pretty much a win-win: both companies have a lot of clout, and I think they’ll both benefit from each other’s expertise and technology. Google needs SAP’s enterprise credibility and customers and SAP needs Google’s AI and IaaS capabilities. You can read more about the collaboration here.
Another big story revealed last Wednesday was that Google had acquired Kaggle, the world’s biggest community of machine learning enthusiasts and data scientists, with the stated purpose of further “democratizing artificial intelligence”. There’s that “democratizing” word again. Google says it will give the community the ability store and query huge datasets, and together the two will continue to support AI training and deployment services.
To be fully successful, Google will need to find a way to not dominate the once independent group, maybe in a way they manage a developer program. They’ll need to resist the temptation to push their own AI frameworks and services or it could backfire on them.
Introducing Engineer-to-Engineer support
Google admitted at the show, which I appreciated, that they under-scoped the need for such comprehensive enterprise services when they launched GCP. Google Cloud lead and SVP Diane Green said services has been one of her largest areas of people investment since she came on board and it’s a smart investment.
Google announced that they would be introducing a new, role-based subscription model of Engineering Support, which comes in three different flavors: Development, designed for developers or QA engineers that can settle for a response time between 4-8 business hours (priced at $100 a user per month), Production, with a one-hour response time (at $250 a user per month), and On-Call, which guarantees a 15-minute response time from a Google engineer, 24/7 (at $1,500 a user per month).
These are all part of a bigger picture Google is embracing when it comes to support for the Google Cloud Platform (GCP)—transitioning from the traditional customer/vendor model, to one more like a partnership. This approach began last year with Google’s launch of their Customer Reliability Engineering (CRE) program, and on Day 1 of the event, they announced their first partner in the CRE program, Pivotal Cloud Foundry. Together, the two companies will collaborate with the goal of improving overall reliability for customers who choose to run Pivotal Cloud Foundry on the GCP. Google also announced that they are partnering with Rackspace with its Fanatical Support to offer managed support for their joint customers on the GCP. Adding the Rackspace capability in a big win for Google.
Google in schools
Google proudly announced that Chromebooks were the number one selling device amongst laptops and tablets in Swedish educational institutions for the year of 2016 (citing FutureSource). It seemed a bit of an odd case study until I did a little research. To some (not me), Sweden is a canary in the coalmine when it comes to education technology adoption.
Citing that same study, all told, there are an estimated 20 million students worldwide using Chromebooks and Google Classroom, and an estimated 70 million are currently using G Suite for Education. My kid’s middle and high school use G Suite and when asked, the administration says they use it because it’s basically free to them and they tell me it’s easy to manage. I’d expect a Microsoft response soon, which I think will be a real battleground to come.
Google also announced that in addition to the $3 million in free credits they recently granted to National Science Foundation BIGDATA research project, they’d also be expanding the Google Cloud Platform Education Grants to include 30 countries worldwide. These grants are a great resource to educators, providing free credits so students can access tools like the Google Cloud Platform. It’s nice to see Google investing in their future like this—after all, the bet here is to get 100s of millions of young users using Google before they get to jobs where they may be using Microsoft Office 365 or Cisco Spark.
Pairing Google Cloud with Chrome and Android
Though not technically an announcement, Google did exhibit some of the ways they were using Chrome and Android to engage with customers and improve worker productivity. Coca-Cola has designed Chrome-based digital signs for grocery stores, using Chromebit devices connected via Google Cloud Platform. Easily managed, these smart signs can use sensors to give the company useful information through Google Analytics. They can also deliver advertising to other nearby screens like customer smartphones. If you’re wondering what the supermarket of tomorrow will look like, I’ve got a feeling we’re going to be seeing a lot more things like this pop-up.
Another example given was Rentokil Initial, a pest control company. Using Android Devices, and by leveraging Google Cloud machine learning capabilities such as the Vision API image classification mentioned earlier, they’re making it easier to identify pests and find treatment solutions. I love this use case as it can be translated to so many like a warehouse pick-pack and shipper. In the realm of manufacturing, Google offered up the example of 42Q, a product division of the manufacturing service provider Sanmina. They built what they call a Manufacturing Execution Systems solution on the Google Cloud Platform, which with the help of Android and Chrome devices is bringing more transparency to their customers’ operations.
Clearly, there is a lot of new stuff to chew on here from Google Cloud Next ‘17—and this only covers Day 1. Ironically, even though Google is a technology and consumer cloud and services powerhouse, I believe they have an uphill battle with enterprises who aren’t cloud-native.
First, all that great consumer technology needs to be made “enterprise-worthy” and even add enterprise-only technology and services. They’ve done a lot of this already. Many enterprises I talk to are still spooked on the security, data locality and control, which may be fair or unfair depending on what angle you look at. Then Google needs to partner to fill the gaps and as the company put it, “meet them where they are”, even if that means partnering with SAP and a company’s favorite programming environment even if it’s from Microsoft. They are doing a lot of that, too. Then Google needs to speak in a way that enterprises, channel and the ecosystem understand. This was part of what the event was intended to do. The industry analyst event I attended was indeed impressive as they provided an early look into their future and access to nearly any executive.
Google CEO Sundar Pichai did a strategic redirect last year when he declared Google as an “AI company” versus mobile. It’s great to see all the great work Google has been doing in the realm of machine learning in the cloud—that’s an industry that’s really blowing up, and Google’s right there in the middle of it. I think their new model of Engineer-to-Engineer support looks promising and I can see enterprise IT’s desire to directly collaborate with Google engineers as no one questions Google’s engineering prowess. Google will need to deeply understand the enterprise environment as well as their own to pull this off successfully.
It wasn’t evident to me after day 1 why enterprises should choose Google over Microsoft, Amazon.com AWS or IBM Cloud for that matter. I don’t know if this was just “leadership-style communication” where you don’t compare yourself to competitors, but you really had to dig deep for differentiation. Maybe it’s machine learning in the cloud, maybe it’s tech prowess, it’s hard to say at this point. Alphabet chairman Eric Schmidt did confidently tell the audience that Google “has both the money, the means, and the commitment to pull off a new platform for computation globally for everyone who needs it.” From a commitment aspect, I think Google would help sway the naysayers by offering some kind of longevity guarantee around their services if they did get cold feet and withdraw. Everyone still needs to respect that Alphabet and Google still derive most all their profit dollars from advertising, a far cry from enterprise services. This is still on IT’s mind, fairly or not.
Stay tuned for our analysis of the content that came out of Day 2—I’m sure we’ll have plenty more to discuss.