Updated on April 28, 2017 by UpGuard
With the introduction of Google’s IaaS dubbed Compute Engine, more than one pundit has declared that Amazon’s EC2 giant has finally met its match. (Not that other vendors don't have their own benefits; check out our comparisons — AWS vs. DigitalOcean, Rackspace, and Azure.) It’s true that Google is one of the few companies that can seriously go head-to-head with Amazon’s and match its computing capacity and prices. But in a significant number of other ways, Google Compute Engine (GCE) is also quite different from Amazon’s EC2. Let’s take a closer look at the two.
GCE is Google’s IaaS offering in the increasingly crowded cloud computing market. It was launched in mid-2012 and started operations proper in April 2013. GCE utilizes a novel billing model that differs somewhat from the norm and is sure to have users of other IaaS vendors why they aren’t getting the same deal. As per Wikipedia’s article: “Each VM is charged for a minimum of [an initial] 10 minutes followed by 1 minute increments, rounded up to the nearest minute. VMs can be launched from the standard images or custom images created by users.” To sweeten the deal, Google also throws in sustained-use pricing, in which the hourly rate is tiered downwards the more you use the virtual machine in a given month. At 100% usage in a month for example, a customer will have saved 30% on the baseline rates.
GCE offers the standard array of features of a fully-fledged IaaS setup: Windows and Linux instances, RESTful API’s, load balancing, data storage and networking, CLI and GUI interfaces, and easy scaling. GCE servers are distributed in 3 zones around the world – central USA, Western Europe and East Asia. GCE offers real 100% uptime via transparent maintenance. If the machine running your application has a hardware failure or needs a software patch, your instance is automatically migrated to another server.
When comparing cloud computing providers, the price-for-performance comparison is critically important. A caveat though – it is almost impossible to do an exact apples-to-apples comparison because of the differing specs of machine configurations offered by different vendors. That said though, GCE still emerges as the victor in this battle.
First off, Google’s instances start up much faster than Amazon’s, by a factor about 60-80%. And in a DaCapo benchmark suite run in March 2014, GCE instances completed the 14-test suite in a score of 575 seconds compared to Amazon’s 719 seconds. In fact a Google machine had the fastest time in 13 of the 14 tests. Price comparisons were also very favorable for Google. At the entry level, an Amazon m1.medium instance (1 CPU, 3.75 GB RAM) cost US$0.12/hour while a similar GCE n1-standard1 instance (1 CPU, 3.75 GB RAM) cost US$0.10/hour. At the high end, an Amazon m3.2xlarge VM (8 cores, 30GB RAM) cost US$0.90/hour, while the similar GCE n1-standard8 cost US$0.83/hour (Source).
And these prices are almost guaranteed to reduce very soon. Cloud computing companies are currently locked in a vicious tit-for-tat price reduction war in which they routinely slash prices a few times a year. The main players in this fight are Amazon, Google and Microsoft, and those of us who use their services cheer on gleefully as we all benefit.
Amazon is of course the biggest and oldest shark in the tank when it comes to cloud computing. Founded in 2006, Amazon’s AWS has about the same computing capacity as the next 12 vendors combined. So nobody, not even fellow corporate titans Google and Microsoft, can really compete with Amazon on capacity. And not many can also compete on the breadth of offerings in the Amazon AWS stable – storage, compute, databases, networking, configuration management, content delivery and many more.
Amazon also owns the largest data centers in the world, and here is one score in which it handily beats Google. AWS data centers are located in 9 regions around the world: 3 in the US and the other 6 scattered strategically around the world. A tenth one is coming up in China exclusively for that market, and will be completely isolated from the others to placate Chinese concerns about US government snooping. In 2013 Amazon also won a coveted contract to create GovCloud, a private cloud expressly for the U.S. government. Clearly, Amazon is the 800-lb gorilla when it comes to cloud service providers-- because of this, we've created a 2-part post on implementing GuardRail with Amazon AWS EC2.
Although we have focused here on compute facilities, these vendors are giants that easily offer a huge array of other services. From NetworkWorld: “Not only does AWS have EC2, the direct competitor to GCE, but it also has a variety of storage options (like Google), including Simple Storage Service (S3), Elastic Block Storage (EBS), and RedShift the data warehousing platform.
The companies each have data analytics platforms too. While Google has BigQuery, AWS answers with ElasticMap Reduce, its Hadoop on-demand service. AWS at its re:Invent show rolled out Kenesis, its real-time data analytics platform for short-term analytics jobs.
Where AWS clearly has an edge over Google is in the marketplace of partners that run services on top of AWS’s infrastructure. In the AWS marketplace users can spin up hundreds of types of services, like enterprise applications from SAP, or performance optimization tools from Riverbed and others. (Source)
The most important IaaS components are computing power and storage. Google’s current offerings in this space are both cheaper and better-performing than Amazon’s, with the cherry on top being a refreshingly simpler pricing structure that gets cheaper as you use it more. But Amazon has been at it much longer, and has many more data centers throughout the world. Amazon also offers a rich plethora of features apart from the core compute and storage, though Google matches it in some areas and will undoubtedly add more as its own platform matures.
For an overview and comparison of the leading cloud service providers, including Google Compute Engine and Amazon AWS EC2, check out our cloud service provider roundup.
Misconfigurations are an internal problem that emanate from within the IT infrastructure of any enterprise; no hacker is necessary for massive damage to occur to digital systems and stored data. And the problem is pervasive, with Gartner estimating anywhere from 70% to 99% of data breaches result not from external, concerted attacks, but from internal misconfiguration of the affected IT systems.