Google Cloud Platform is an Infrastructure-as-a-Service (IaaS) solution which begins at $6,092 per year for a eight-server Web app package (as described below). Google Cloud Platform is heavily influenced by Linux. For one, at its most fundamental layer, it’s powered by its own Google Linux variant. But, since you’re unlikely to ever work with it at that level, that shouldn’t be cause for concern. In addition to its own Linux flavor, Google Cloud Platform exploits Linux’s KVM hypervisor to offer customers virtual infrastructure running 64-bit CentOS, Debian, Red Hat Enterprise Linux (RHEL), SUSE, and Ubuntu Linux, in addition to Windows Server 2008 and 2012. Its default Linux virtual machine (VM) is Debian.
Google Cloud Platform also offers perhaps the best container support currently available on a public cloud. It supports both Docker and CoreOS container standards. You can manage them with Google’s own Kubernetes container management tool. It’s a good idea to get familiar with Kubernetes prior to pulling the trigger on a Google Cloud Platform contract. Since everything on Google runs on containers and Kubernetes is Google’s own management tool, you’ll likely be using it a lot so make sure you like it. I haven’t tested all container management tools but I liked what I saw of Kubernetes.
Another nice feature of Google Cloud Platform (and this is not surprising when you consider it’s Google) is that sophisticated networking comes standard. You can easily create a virtual network, including private and public subnets on which to arrange your VMs. In addition, firewalls, routers, and gateways are simple to set up—as long as you’re comfortable with the Linux command line. While Google does offer a Web-based management guide, many of the tutorial explanations in Google Cloud Platform’s online docs refer only to command-line execution. This is in keeping with other IaaS competitors, including our Editors’ Choice winner Amazon Web Services$6,415.00 at Amazon as well as IBM/SoftLayer $9,100.00 at SoftLayer, that are targeting customers with real IT administration experience, not simply power users. As long you fall into that group, Google Cloud Platform also makes it simple to assign temporary IP addresses aka Ephemeral IP to reserved IP addresses.
Setup and Configuration
As mentioned, however, for straight IaaS management and configuration tasks, Google Cloud Platform’s online documentation often refers users to the Google command line, which amounts to a Linux variant. But, for basic setup and management tasks, the Google Cloud Platform’s Web-based tools work well. For example, I used the Google Developer Console to easily create a simple Web server app. This used a Debian Linux VM with a MongoDB database engine using a single vCPU, a 10GB SSD, and 3.75GB of RAM. I did this with only a few clicks.
There are other alternatives, however. For example, spinning up my Debian VM with its MongoDB database engine could have been handled through Google’s wide range of prepackaged server software stacks. For example, I spun up a SugarCRM instance in less than 15 minutes. The only work I had to do was to click the application icon from the Google Cloud Platform Cloud Launcher.
Of course, it helps if you can SSH your way into the VM but, for basic functionality, Google Cloud Platform will let even the most IT-anemic small to midsize business (SMB) start up common business servers. However, once you eek past basic functionality, Google Cloud Platform will quickly route you to its command-line standard so, unless you’re interested in only single or minimal server installations, an IT background will most likely be required.
Google Cloud Platform Performance
You would think that with Google running things, Google Cloud Platform would have data centers everywhere. But you’d be wrong because they don’t. Instead, there are only three sites: one in the United States (in Iowa), one in Belgium, and one in Taiwan. Of these sites, the Taiwanese one currently only has Intel Ivy Bridge processors powering its bare-metal hosts. The others go all the way up to Haswell. Google has opened a data center site in South Carolina, which will give Google Cloud Platform an American East Coast presence.
However, that still leaves a lot of open territory. And it means that, unless your business is close to one of these primary sites, Internet performance may slow your overall Google Cloud Platform performance. Just like any other computing service, whatever your slowest link between you and your resources is determines your user performance experience.
Within the Google Platform Network itself, network performance seems excellent. I say “seems” because it’s hard to tell from the outside looking in. Google only uses 10GB Ethernet within its cloud data centers but I strongly suspect that Google uses its content delivery network (CDN) for cross-data center communications (unlike, say, Editors’ Choice winner Amazon Web Services, which uses the ordinary Internet). That said, if you’re not in the U.S. or the European Union (EU), you can probably find better performance with another cloud provider.
Testing Google Cloud Platform
As for the service itself, to test it I used cross-platform processor benchmark Geekbench 3 from Primatelabs. For Google Cloud Platform, I used the minimum configuration needed for the benchmark: an n1-standard-1. This platform consists of a single Ivy Bridge Virtual CPU (vCPU), 3.75GB of RAM, and a 10GB HDD. The benchmark instance consisted of 64-bit Ubuntu Linux 14.04 Long Term Support (LTS), with Python 2.7 (which ran the test for 24 hours).
This benchmark program runs several integer, floating point, and memory tests. With this benchmark, higher numbers are better. I didn’t try to formally test networking or storage performance. Cloudlook reports, however, that the n1standard-1 primary disk rate averaged 52Mbps.
Google Cloud Platform delivered the best benchmark numbers of any competitor, outside of the performance of Rackspace Managed Cloud$10,300.00 at Rackspace, that is. Rackspace Managed Cloud, however, needed two vCPUs to achieve its mark, while Google Cloud Platform only used one. That means, per processor, Google Cloud Platform is the fastest cloud service I tested in this IaaS solution review roundup, with a score of 2,558.
Pricing and Contract
Google Compute Engine’s service level agreement (SLA) guarantees at least 99.95 percent uptime. If your monthly uptime percentage is between 99.00-99.95 percent, you’ll get a 10 percent credit is received. If it drops to between 95.00-99.00 percet, you’ll get 25 percent. If it goes below 95.00 percent in a month, you’ll get a 50 percent credit. While that’s a similar structure to many of the SLAs represented in this IaaS solution review roundup, those are currently the best numbers in the field.
One nice feature Google offers is a Pricing Calculator, which seems more accurate than most calculators. For example, were I to run the benchmark for an entire month, defined by Google as 720 hours, it would have cost me $32.79 a month. Your cloud, I can guarantee, will cost you more. All in all, though, I am certain you will get great performance for a good price from Google Cloud Platform.
That said, at $32.79 per month for the benchmarked instance, Google Cloud Platform was one of the most affordable services in this review roundup. Google is fighting a price war right now with Amazon Web Services and Microsoft AzureLearn More at Microsoft so your bill may differ significantly from ours, even in the very near future.
So much for the theory, let’s look at the practice. I used RightScale’s Plan for Cloud Calculator to price out a three-tier Ubuntu Linux Web application. This was made up of two on-demand micro instances: the load balancer and the hosting website. The Web server itself was hosted on a g1-small Linux instance. These were supported by a pair of on-demand, f1-micro Linux Web servers for peak demand and a g1-small Linux disaster recovery (DR) server.
For the database backup, I used 300GB and 4GB storage residing on Google Cloud Storage Standard. The 150GB DBMS itself resided on a Cloud SQL D8 MySQL instance. The DR DBMS with 300GB of storage made its home on a Cloud SQL D2 MySQL. The total monthly data transfer allowance, source to destination, was 440GB, with a destination to source allowance of 1,140GB.
As for tech support, I went with the minimum available support system. This help desk provided assistance via both email and tickets. In total, this simple eight-server Web application package costs about $6,092 per year.
Google Cloud Platform is affordable and the second-fastest service in this review roundup. When compared to Microsoft Azure, I rank them about the same. At day’s end, if you need to choose between the two, I’d go with Google Cloud Platform if I were running Linux (especially with containers), and with Microsoft Azure if Windows was my VM of choice.