Google announced on Aug. 23 a new two-tier network service model for customers of its cloud services.
One is a so-called Premium Tier for organizations that want a high-performance, low-latency cloud infrastructure for running their workloads; the other is a Standard Tier for more cost-conscious organizations.
The Premium Tier is what organizations get by default and is what Google currently offers all of its cloud customers. With the premium service, all traffic from enterprise endpoints to Google’s cloud hosted apps travels over the same network that Google uses to deliver its own services such as Gmail, Search and YouTube.
The private fiber network currently has more than 100 points of presence (POP) around the world. When an employee at a company that has signed up for the service accesses a Google cloud app, traffic from the user’s end point is routed to the closest POP from where it enters the Google network. Traffic from the application back to the end user is routed in the same manner to the POP closest to the user and delivered from there.
This ensures that in most cases, traffic to and from Google reaches its distance in one hop, thereby ensuring low latency and high performance, said Prajakta Joshi, product manager of cloud networking at Google, in a blog announcing the new tiered cloud service model.
Premium Tier customers will get other benefits as well, including better service resilience and availability than the standard service. For example, Google currently maintains a minimum of three separate and independent paths for routing traffic between any two Google network locations. This ensures that traffic can flow continuously between Google and the endpoint even if one route—and in many cases two routes—gets disrupted for any reason, Joshi said.
In addition, businesses with the premium Google service will get support for loading balancing client traffic across both IPv4 and IPv6 networks globally.
The lower cost new Standard Tier is designed for organizations with less stringent performance and latency requirements. The main difference between the Standard and Premium tiers is that with the former, network to and from the user endpoint is routed through transit networks owned by third-party ISPs.
When end users access a Google cloud hosted application, the traffic first is routed over ISP networks and enters Google’s private network only in the region where the app is hosted.
According to Joshi, performance on Google’s Standard Tier will be comparable to what other cloud services provide but will be lower than that offered by the company’s premium service. Service availability and redundancy will depend to a large extent on the ISP’s network, and customers can expect some congestion or outages, she said.
Pricing for the services will continue to be based on the volume of outbound traffic from Google’s cloud platform to the internet. Google cloud customers in North America and the European Union who are on the Premium Tier will be charged $0.105 per gigabyte of traffic for the first 10TB and lower amounts as traffic volume increases. In contrast, pricing for the standard service in these two regions starts at $0.085 per GB for the first 10TB and then drops down for bigger traffic volumes.