I’ve just spent a few days at the extravaganza which is Google IO. As with previous occasions, this year’s event didn’t fail to blow away the attendees with the sheer amount of “wow” that Google manages to pull out – on so many fronts Google is doing incredible stuff. My main interest however, beyond the bling, was to look at the announcements around the Google Compute Engine and Google App Engine – together the two public cloud infrastructure initiatives that Google has. The usual disclosure needs to be given however, Google comp’d my entry to the event and, along with all other attendees, I was given a Pixel. Before the analysis comes the high level news:

  • Google Compute Engine is now in general availability
  • GCE offers sub-hour billing charges for instances in one-minute increments with a ten-minute minimum
  • Shared-core instances provide smaller instance shapes for low-intensity workloads
  • Persistent disks support up to 10 terabytes per volume
  • Google App Engine adds PHP support
  • GCE, GCS and GAE all achieve ISO 27001 certification
  • Google launches Cloud Datastore, a schema-less solution for storing non-relational data

Right, on to the insight. We’ve been waiting a long time for a real competitor to AWS to appear. Even Adrian Cockcroft, Chief Cloud Architect for Netflix and perhaps the biggest platform evangelist for AWS (either inside or outside of AWS) has stated that he’d love to see some credible alternatives to the Seattle behemoth to appear. As OpenStack moves more and more into a private cloud play, and Microsoft continues to confuse the world about exactly what its cloud platform is, those options have been sparse (actually, non existent).

That could change very quickly now as Google sees to be diverting some considerable attention to it’s compute platform. As my buddy Alex Williams pointed out, Google Compute Engine was one of the top three areas for attention on the IO agenda – that is indicative of Google’s strategy and is, at least in part, an affirmation that the Goog has been looking closely at the massive adoption that AWS has. Not so much because of the revenue it can drive (although that too) but more because whoever deliver the platform chosen by developers is well positioned to capture the bulk of the highly valuable user data that Google has so effectively monetized.

It was interesting to contrast the attitudes towards GCE that were evident at Cloud 2020, the cloud summit myself and colleague Krishnan Subramanian put on in Vegas the week before IO. There were two distinct opinions raised at that event:

  • Google has the scale and experience to be a real competitor to AWS in the public cloud space
  • Google is a high margin business, which Amazon is in the low margin one. As such Google’s new found love affair with the public cloud will soon wane

I think it’s short sighted to write off Google merely from a first degree margin perspective Google of all organizations knows the value of secondary monetization of user data and has built a business out of giving away commodity services in order to mine and leverage the data it can extract from those services. GCE is hardly likely to be significantly different. if we accept that Google is committed to GCE, then it’s time to look at what sort of growth expectation that platform is likely to gain.

Massive – quite simply AWS should be very concerned about this news for two distinct reasons.

Firstly – Google has arguably the best experience of any organization on earth at building massive scale data centers, networking elements and distribution networks. They deliver incredible scale computation on a global basis with high degrees of reliability. With all of that plumbing in place, it’s not surprising that the past few months have seen Google very quickly expose functionality to customers. I’ll say it again, the functionality is all there, Google now just needs to expose the APIs to allow people to use it. The fact that Google has little to build beyond the endpoints means that the relative pace at which they can roll out functionality is higher than its competitors.

Secondly – Developers are increasingly wishing to make use of the APIs that surround the data and services that Google holds. Given the option of using a public cloud provider whose infrastructure sits right next to that data, it makes sense for developers to use Google as their infrastructure of choice – it’s a nice solution to at least part of the data gravity problem that developers face. Add to that the release of the higher-value products that Google offers (Android Studio for example) and there’s some extra value built in to the Google platform that AWS can’t replicate.

During IO I took the opportunity to talk with some developers and ISVs who have been using the Google Cloud Platform – they report a truly amazing level of performance. In fact one person I talked with suggested that the performance for his management solution was better than if he was using physical hardware with storage and compute in the same rack. This performance is a function of the amazingly performant network that Google has built over the years.

So. Is GCE a complete slam dunk? Not yet. Google has yet to prove that it really understands enterprise needs, and it needs to prove it’s in this business for the long haul. The fact that Google Apps, for example, was hardly even mentioned at IO is an indication that this is a company that has a fairly relaxed attitude towards what business really needs. Google needs, if they’re serious about being a credible IaaS provider for larger workloads, to build out the non technology aspects of a vendor – on that count we’re yet to see what will happen.

That said there was an interesting announcement at the tail end of IO on Friday. Details are scant but apparently Google showcased some new networking capabilities that enable hybrid clouds to run between GCE deployments and on-premise data centers. The functionality allows customers to establish virtual private Layer 3 networks and assign static IP addresses to instances. Apparently there is the ability to connect networks and run load balancing across them. Details are scant but it’s an unusual, and surprisingly business-focused offering.

Frankly having AWS as the kind of the IaaS castle has been boring. All things being equal, Google announced to the world last week that it’s committed to storming the castle. Time will tell if it succeeds, or drowns in the moat.

Ben Kepes

Ben Kepes is a technology evangelist, an investor, a commentator and a business adviser. Ben covers the convergence of technology, mobile, ubiquity and agility, all enabled by the Cloud. His areas of interest extend to enterprise software, software integration, financial/accounting software, platforms and infrastructure as well as articulating technology simply for everyday users.

  • re: the second opinion above: if google was worried about low margin businesses… youtube would have been shut down long ago. it is one of the rare long term innovators in infrastructure.

  • Well, as a Chromebook user I’d certainly like someone to hand me a Pixel as I walked through the door to Google IO. I don’t think AWS would do the same. That said, who else besides Google is in a position to offer AWS some competition? Microsoft and Rackspace are long shots. You are right to point out that one of Google’s strengths is its OpenFlow optimized G-Scale network that connect its data centers with 10GbE links that operate at close to 100% utilization. Google also owns tons of fiber-optic networks. If Google was an ISP in would be the second largest behind Level 3. Yes, there is always the BIG question about whether Google is serious about delivering services to enterprise customers. If Compute Engine and App Engine become significantly better offerings than AWS, people will use them for the reasons you mentioned. AWS defined Cloud 1.0 and Google is defining Cloud 2.0. Yes, it will be interesting to see how it plays out.

    • Not sure how 100% utilization on their network is a strong selling point… 🙂

      I would rather that they have tons of excess compute and bandwidth capacity, and leverage that to undercut the AWS pricing models.

      • Google’s G-Scale network utilization could be interesting to GAE or GCE users deploying to multiple Google data centers. High G-Scale network utilization is also an advantage over Google’s competitors, who may only be able to achieve 30 to 40 percent utilization on their 10GbE data center network links.

  • As a side, interested to know why you think microsofts public cloud is confusing? Id say they are up there competing with aws and rackspace more than google?

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.