Cloud

TCO of Cloud Computing

TCO of Cloud Computing

One of the most difficult items to forecast for new IT investment is the Total Cost of Ownership (TCO) model.  Calculating TCO was never easy for purchased on-premise software or hardware packages, but it is more complicated for cloud computing solutions. You’ll hear various vendors talk about turning capital expense into operating expense, where on-premise software is a capital expense purchase, and cloud computing is an operating expense.  However, this is overly simplistic. To truly evaluate the TCO of a cloud computing investment you need to look at the following categories of spending.

Cloud licensing
Since cloud licenses are like rental fees, you will pay more as you use more of the cloud service and, therefore, your TCO model needs to factor in a growth model that is realistic over the time horizon of your TCO analysis.  Usually your business partners can tell you their forecasted growth percentages and you can utilize these to mimic your cloud growth percentages.

Most people don’t predict to shrink, but you still need to negotiate the ability to reduce your cloud licensing if your company downsizes. If you don’t, you risk your TCO model being wildly off when growth is replaced with downsizing.

Cloud integration
For those of us who have already implemented SOA/Web Services for our integration architecture, integrating with the cloud is not difficult.  However, be prepared to spend time building more integrations than you originally expected. As your users get used to working in the cloud, they will want the cloud environment integrated seamlessly into your on-premise software or hardware world. Planning for that expense will serve you well. Also, don’t forget to plan costs for additional capacity and scalability in your integration environments. Remember, your integration environment won’t get smaller with cloud computing, they will get bigger and more complex.

Cloud monitoring
Having a cloud vendor doesn’t mean you should stop monitoring whether you are meeting service level agreements (SLA). It is unlikely the vendor will have the exact SLA measurement or monitoring that you need. Therefore, you will need to use your own monitoring solutions to build things such as script monitoring, synthetic transaction monitoring, systems management tools monitoring, and extraction of cloud logs into your internal log management system.  All these items take time and effort to setup, test, debug and train your Network Operations Center (NOC) or Security Operations Center (SOC) to manage.

Cloud security
Most of us have invested in Single Sign On (SSO) and Identity Management (IDM) solutions that work well with our on premise environments. Going to the cloud means another SSO setup and changes in how your IDM solution is architected for on-boarding and off-boarding. Plan ahead for making these changes.  Your users will expect integration with SSO, and your security and audit teams will want IDM integration in place, especially if you are in a regulated industry or publicly traded company that requires tight controls around user access.

If you plan on storing secure, personal, or private information in the cloud, you must be prepared to incur expenses conducting vulnerability testing of your cloud provider, exactly as you would internally with an on-premise system. Unless you can get the provider to send you vulnerability reports on regular basis, preferably monthly, you’ll be best served by doing your own vulnerability testing. You’ll also need to negotiate a SLA for how quickly the vendor has to address any issues discovered during monthly scans.

Refresh savings
A big win for cloud computing is the elimination of software and hardware upgrade costs.  Make sure your TCO includes cost savings for not having to buy hardware for Production, QA/Staging, Test, Development and Disaster Recovery environments.  This is the biggest TCO win for cloud computing so, make sure you cover all the costs and include your 3-4 year refresh cycle cost savings. Finally, don’t forget to include cost savings for not having racks, cooling, and switching hardware.

Labor savings
A softer cost to factor into your TCO is the reduction in software/hardware upgrade labor costs from not having to do IT testing, user acceptance testing, and IT development effort to re-apply past customizations.  If your cloud offering is a true multi-tenant architecture environment (as opposed to hosting, which is not Cloud), then your upgrades will be seamless and you should have a significant labor cost reduction for eliminating these activities. Make sure your business and IT people understand the positive ramifications of this change…namely that you will not have to set aside time for everyone to help test these upgrades and patches.  Instead you can focus on rolling out new features and dealing with training and change management.  Enabling new features, conducting training and change management, are much more fun to focus on than trying to verify nothing breaks with the latest vendor upgrade just so you can stay current with maintenance.

Lastly, with most cloud computing solutions, if you are moving from an on-premise solution to cloud computing, you will likely have internal labor savings due to reduced headcount. Eliminating the hardware environments and reducing the customization programming required results in reduced labor for engineering, programming and potentially management. Be sure to factor in increases in labor for integration environments/coding, and possibly security/monitoring, but the net result usually is an opportunity to reduce IT headcount. This labor savings makes sense since cloud computing is akin to outsourcing, whereby the cloud provider is providing these labor services and the cost is shared by you and others who subscribe.

Summary
Overall I have not found cloud computing to be less costly than on-premise computing from an out-of-pocket expense perspective. However, I have found that it can be comparable in costs and less costly over a 10 year TCO time horizon.  More importantly, cloud computing can free up significant internal technology and business personnel to focus on taking advantage of new features instead of regression testing and re-applying of customizations to the latest software upgrade, required hardware refresh, or required patch. That is priceless.

Contributor

Eric Dirst, Senior Vice President and CIO, DeVry Inc.

Eric Dirst is Senior Vice President and Chief Information Officer for DeVry Inc. DeVry Inc. is a publicly-held, global provider of educational services and is the parent organization of Advanced Academics, Apollo College, Becker Professiona... More   View all posts

Advertisement

RedMane_Expand_Your_Value_MPU

Newsletter

Login


Not Member Yet? Register

Register

  • Minimum length of 8 characters