Public vs Private vs Hybrid Cloud

This week’s lecture focused on cloud computing, and there are many important classifications within cloud computing. Compute as a service is the best way to describe cloud offerings, and there are three main categories currently: infrastructure as a service (IaaS), platform as a service (PaaS), and software as a service (SaaS). Another consideration for cloud compute that both  both Carl Eschenbach from Sequoia Capital and Jeff Welser from IBM Research spoke about is public vs private cloud. Recently, there has been a push for hybrid cloud as well. Before this lecture, I was unfamiliar with many of the distinctions between public vs private vs hybrid cloud, but the information presented in this week’s lecture coupled with my own research allowed me to better understand the current trends in cloud compute.

Public vs private cloud:

Public cloud offerings are relatively well known, especially from the main providers such as Amazon Web Services (AWS), Google Compute Engine, and Microsoft Azure. These services can be accessed by anyone, and compute infrastructure is shared across all users, which amortizes the cost of physical hardware for cloud providers. However, this could pose a problem for enterprise use because data must be secure, and shared compute resources could bottleneck important processes. Both public and private cloud deployments have their own advantages and disadvantages, which has caused a drawn out debate of which deployment to use. An article I found on networkworld.com summarizes the pros and cons of each. “Private clouds come with an inherent control of the infrastructure, which gives some security-conscious and regulatory-sensitive customers peace of mind. On the other hand, true private clouds that allow for virtualized resources to be self-provisioned by users can be difficult to set up and manage.” [http://www.networkworld.com/article/3131942/cloud-computing/which-is-cheaper-public-or-private-clouds.html]. Public clouds have the advantage of access to “an almost infinite amount of infrastructure resources without any upfront investment required” along with the “ability to use cutting-edge technology available first from public cloud vendors” (networkworld.com). Access to the latest technology from public cloud vendors was brought up in lecture multiple times as well, such as Amazon’s Marketplace for applications, so this does appear to be a significant advantage for public cloud deployment. TCO, or total cost of ownership, is mentioned many times in this article, and it is the most important metric for datacenters and cloud deployment. With the proper utilization, efficient management, and appropriate deployment size, having a private cloud could be more financially viable and have a lower TCO. This debate does not appear to have and end in sight, however, hybrid cloud deployment is a recent trend that provides a compromise between public and private cloud.

Hybrid cloud:

While the use of public and private cloud will not end anytime soon, an article from wired.com states, “increasingly the best solution for many clients will be a hybrid cloud approach. By developing the ability to mix and match the best elements of private and public clouds, a company can evolve its existing IT assets” [https://www.wired.com/insights/2014/10/public-vs-private-cloud/]. To better understand hybrid cloud, I read an article from zdnet.com which describes hybrid cloud as a combination of a public cloud offering with private cloud infrastructure, and this allows both ends to communicate via a fully encrypted channel. Sensitive data would be stored entirely on the private cloud, and the article stresses that the private and public cloud are two entirely separate entities, which allows sensitive data to remain secure, but the encrypted communication channel allows for heavy computations to be completed on the public cloud. [http://www.zdnet.com/article/hybrid-cloud-what-it-is-why-it-matters/]. Another benefit includes low latency access to the private data, since infrastructure would be on-site and data would not need to go through the public internet unless the need for more compute resources arises, in which case the public cloud can be used to offload the workload. Having intermittent access to these compute resources also lowers cost, since there is a pay-per-use model. As with any model, there are also drawbacks to the hybrid approach. If security is very important, introducing a connection between the public and private cloud introduces a vulnerability even with a secure and encrypted channel. If latency is critical, transporting data for compute would introduce a delay that could be unacceptable depending on the application. The example given in the article is a supercomputer designed by Hitachi that determines if a tsunami warning should be issued following an earthquake, and this is extremely time-sensitive, so offloading compute to a public cloud would not be feasible.

Conclusion:

No clear winner seems to have emerged between public and private cloud and the debate continues on today, but a hybrid deployment could provide the best of both worlds. Data security offered by private cloud is crucial, but access to the latest technology is only feasible with public cloud. Thinking into the future, the importance of access to the latest technology will only become more important. As Jeff mentioned towards the end of lecture, IBM now offers quantum compute via their public cloud and users have access to 16 q-bit currently. Jeff mentioned that when 50 q-bit and beyond becomes accessible, it will open up applications which are currently impossible. Having access to this sort of cutting edge compute as soon as it is available will allow us to solve some of the world’s most important challenges.

Users who have LIKED this post:

• 
•