The State Of Cloud Computing
Cloud computing is yet another buzzword that has been metamorphosing constantly. In other words, it continues to grow but the evolution has been very zigzag. A few decades ago, IBM used to offer its mainframes to banks, utilities and other industries on a rental basis and the billing used to be metered. Even today IBM is doing the same thing with much success thanks to the amazing performance of its mainframe servers. However there are various new forms of cloud computing that have made it a common household word.
Salesforce is the company that showcased to the world more than a decade ago what SaaS could do. Having said that, SaaS, with multi-tenancy and metered billing model at the heart of it, is more applicable for product and services companies that offer per use services over the net. A lot of verticals, for better or worse, are not comfortable with the fact that the data lives somewhere outside of their brick and mortar structure. These fears in the minds of the end user companies take us to the other forms of cloud computing i.e. Paas, Iaas, Hybrid. In short, very little overall IT cost incurred, good service availability and a per use model for user companies is what SaaS brought to the table.
PaaS provides all kinds of hardware, software stack and tools to enable an IT department to deploy apps over a cloud. However PaaS seems to be lost when compared with the other two flavors i.e. SaaS and IaaS. Developers have to deal with myriad version mismatches of Java, orPHP or Microsoft libraries and frameworks and often work around them to push something into production. Moreover to make matters worse, PaaS is expected to keep up with the dynamism of IT industry i.e. keep adding newer shining frameworks and libraries. For instance, if there is library that helps plumbing social media integration or improve mobile compatibility or provide MBaaS, then a PaaS framework has to get it sooner than later in order to stay relevant and retain customer base. Add to that the cost incurred when an end user type of customer buys a piece of cloud infrastructure and takes some time before actually getting it to work for it. In short, PaaS has been struggling to make its mark and leaves much to be desired.
IaaS involves providing servers and VMs and then using them to build application architecture totally to your own likes with little or no restrictions. This is very popular amongst verticals that are jittery about their data due to regulatory and privacy concerns. However the hourly billing of server and uptick in price that you have to bear with when you buy 99.9+ availability or fail over or clustering, tends to make IaaS cost prohibitive to have a lot of servers. In short, maintaining an IT or DevOps team wherein servers are built out when an application is ready to go to production becomes a costly option but nonetheless worth the peace of mind that IaaS brings to the table.
This incarnation of cloud computing is better for customers who run massive analytics or batch jobs on several nodes concurrently and find the hourly billing of server provisioning a bit predatory. Hybrid implies using a combination of own servers and cloud servers together to achieve high thorughput.
The Hybrid advantage:
One good way to use hybrid is to start the work on your own servers and perform all the ground work beforehand. Once the batch job or data crunching needs start to become enormous or exponential wherein a whole lot of servers are needed to process the data in a timely fashion, cloud computing kicks in. By dividing the work into a set of logically separate pieces and running them in parallel on several cloud servers, very high throughput can be achieved without spending a lot of dollars on cloud infrastructure. A business criterion for, dividing and running the job on several servers in tandem, needs to be evolved. For starters, you can apply Pareto’s principle (also known as 80-20 rule). This rule dictates that 20% of your customers bring 80% business. These are the customers that deserve provisioning of a separate node for processing their data while rest 80% customers can be processed by a couple of in-house servers.
What is your cloud strategy? What have been your experiences in the cloud space? Share with us.
Take charge of your enterprise app strategy ! Download this eBook with the 10 best practices and get started