Understanding Cloud and Legacy Costs

As I outlined in Part 1 of this blog series, a lot of organisations simply don’t have a detailed understanding of their existing on premises costs. But this is an important part of planning your cloud migration strategy. Because only by gathering information on existing costs, can you begin to identify waste and see where efficiencies can be made. And, indeed where you can earmark funds for future innovation. So how can you quickly get this information? 

Here at Telefónica Tech we use intelligent tooling informed with industry averages to produce a sample TCO (Total Cost of Ownership). 

Gather costs

To help build a business case we’ll capture real figures and feed actual data into a TCO analysis to model against cloud costs.  What’s interesting, is that from this type of analysis we’ve noticed costs for things like power can be uneven across different industries.  

For example, a brewery can pay much less per KWH than other types of companies, including the companies that supply power. Yes, really!  In another example I met an education customer who buys power well in advance from an energy trader to get a discount. While at the other end of the scale, some customers I speak to don’t even consider power part of their Profit and Loss (P&L) statement.  So, as you can see, there is plenty of variety in the approach.

Model costs

This is the fun part! Using our toolkit, we’ll capture a detailed inventory of the on-premises estate and run a performance analysis across a business cycle.

In most cases we use two-or three-weeks’ worth of data. If the customer’s business is known to have a regular spike during the month, we need to capture four weeks (4.3 to be exact) so that the spike does not skew the data.  Once we have a good capture we can model costs.

With the analytics merged with the customer’s Microsoft Licensing agreement, we can present various cost models including correctly sized virtual machines, SQL costs as VM’s, managed instances, DTU or vCore – all with Azure Hybrid Use Benefit (AHUB) and / or reserved instances applied.  It's worth noting, buying reserved capacity and combining this with mobile Microsoft licensing dramatically affects the consumption costs.

We also model the likely cost of ‘sundries’ at this point. Most economic assessments just include compute and storage – and actually, that gets you the right order of magnitude, but we like to paint a more detailed picture that includes things like estimated network and Security Centre costs. With our modelling, we can select/deselect services and even model costs against different Azure datacentres in real-time.

Analysis is purely economic at this stage It doesn’t tell us if an application can run as vCore, just what it will cost to run if it was. Assessing an application for PaaS suitability comes later.

But with this type of reporting you can begin to get a good idea of potential costs.

Assess Cloud Readiness

As part of a general cloud economics assessment, we recommend a parallel workstream to assess in-house applications for cloud readiness.  Customers invariably want to use PaaS where possible, and third-party apps are unfortunately ‘fool’s gold’ in this respect - seldom, if ever, supporting cloud native services in their entirety.  But all is not lost, there is the potential to use native services such Azure Files, load balancing and other components around the 3^rd party app to make better use of Public Cloud. But to really transform the app we must have full access to the source code.

Although the off-the-shelf apps might not support PaaS, the vendor will often point you at a SaaS version of its product which may or may not be of interest.  SaaS is very cost effective as a cloud usage model, as the vendor is on the hook for everything except the data you introduce to it.

But SaaS is also standards based and inflexible. You get what everyone gets and the less well-known down side to SaaS is that it becomes difficult to maintain corporate governance (security) when high numbers of SaaS products are introduced into your workforce.

The biggest opportunities for PaaS are applications that have been developed in-house. For these types of applications, we can scan the code and databases to provide very detailed output to indicate where code must be changed to adopt cloud native services such as SQL DBaaS and Azure App Services. Our analysis will also suggest recommended changes such as adopting SendGrid and Azure key vault.

The detailed scan will display the problematic code, line number and remediation steps, along with the estimated days/hours of effort.  Development Teams find this information a huge help if they are not familiar with native Azure services and concepts.

The information aggregates all the changes and estimates the developer effort in making the required changes based on industry averages. This is useful in the planning stage.

For example, reviewing the detailed reporting in the Cloud Readiness Assessment helped one of our customers to make the decision to move to IaaS in the short term, before looking to remediate applications longer term. It also helped them make the decision to cancel their data centre lease.

If you would like to understand more about a Cloud Economic Assessment or Cloud Readiness Assessment, please get in touch