Cloud on-boarding – Challenges

We spent a good number of years primarily managing the cloud on-boarding of large scale enterprise customers. As a part of on-boarding them to the Microsoft Office 365 Exchange Online SaaS offering, we realized that there are common concerns that keep popping up. Attempt of this first (in the series of three) paper is to de-clutter the space & bring clarity. A basic knowledge of cloud 101 is assumed.

First, let’s start by going over the Cloud stack and models currently in play.

Software as a Service or SaaS Offering as it is commonly called, is the most familiar model, we use this to pre-package Exchange Online as a offering which can be accessed by multiple client endpoints. In this offering, once onboarded, enterprises mostly play with data settings and application configurations but the entire software & hardware deployment is managed by the cloud provider (E.g. Microsoft in this case.)

Infrastructure as a service (IaaS) on the other extreme is a model where the cloud provider (E.g. Msft Azure/GCP/Okta) manages only the network, bare metal servers & storage resources. The customer takes care of the rest of the stack especially the application and data layer. AWS seems to be the strongest player here with GCP catching up relatively fast due to more configurable & lower cost model. This model is less common with SMB’s as they typically prefer not to hire dedicated resources for IT management.

Right in the middle is the PaaS model, where in Cloud provider also manages the OS (VMs + Physical) & the run time environment. Even though this model lowers the total cost of ownership, it is definitely less flexible than IaaS. The advantage here is that, the more time-consuming stuff like OS Patching, deployments and upgrades are all taken care of by the provider which uses the scale to reduce the costs, there by resulting in a win-win.

The image below titled ‘Cloud Models’ beautifully captures the differences – and the key to note here is that ‘Scale reduces cost’ is a model that has been proven time & again, this is also evident by the reduced cost as measured in MIPS (per the Economies of Cloud Scale model).

Other models, which are less common are Data as a Service, Business Process as a Service but at this point for brevity sake, lets skip those jump to the meat !

Why Cloud ?

There are a lot of well researched papers like the one here. In a nutshell, there is a play of the scale here & it serves well to view this from different pivots:

• Large cloud providers are able to drastically reduce costs, by deploying measures like negotiating cheaper labor & electricity costs, more automated deployment and patching mechanisms, supply diversification & such. This can add up to be substantial savings for them and eventually get passed to their customers.
• Second pivot to that is the case for streamlined focus. A food based company like Mondelez, Mars or McDonalds* can better spend their focus and energy on manufacturing and sourcing improvements, instead of managing the servers and their procurement, deployment, maintenance and such.
• A third but perhaps slightly more important pivot to this is their security & privacy concern. With regulations like GDPR (which go into effect 5/25/18 for all EU), DFARs (For Federal Gov & defense subcontractors), NIST & others – Governments and their subsidiaries have made their need for privacy a clear priority. Secure access to data is not an ask any more, it’s an implicit assumption. Top cloud solutions like AWS, GCP, Azure & IBM (and a few others) have two (and even three in some cases.) factor authentication built standard in the service. It is clear to the enterprise CIO’s that they can rely on the certifications (& process assurances) of the cloud providers to ensure safety and privacy of their data.
• Then there are other pivots to this, namely, geo-colocations, data redundancy, network isolation, dedicated CDN End points & such. Those are selectively important for different enterprises depending on their market segments. A cloud provider being elastic by definition can provide all of these; E.g. to meet the geo-colocation needs of an extremely important customers, microsoft opened dedicated datacenters in australia and more recently in South africa.

Cloud consumption

Enterprises consume cloud in a very varied way, but the model essentially remains the same; I am going to save some typing and look for a good picture here. This picture from Interxion does a pretty decent job.

In a nutshell, the core pieces, like identity management, free/busy and inter-organization relation ships, mergers, acquisitions & splits are all managed internally (& typically behind org firewalls and n/w policies). The provider then comes in with SaaS/IaaS or a hybrid custom approach.

In the second article of this series, our team will write about the challenges that an organization faces when on-boarding and some best practices that they can follow to alleviate some of those as well for them to be able to scale for future trends. There is a lot that can be discussed here and I plan to zoom into a few of the areas like pre-onboarding challenges, migrations issues and post-onboarding challenges (yes – there are those as well)**.