If you are considering moving any of your business analytics to the cloud and analytics are something that you believe to be a competitive advantage for your organization, this is where to get started to ensure that continues to be the case.
Cloud analytics are an excellent choice when your use cases fall into the sweet spot for the infrastructure and support required. As an example, if you need short term usage of hardware to move a modest amount of data, provide some computing power on demand and return a series of analytics results or reports to a group of users, you will probably find that some flavor of cloud analytics is right for you. You may also find that cloud analytics work when you need to provide a development platform for your data scientists that is out of the way from your production network and processes.
However, if you have an issue with the security model employed by the provider, you have corporate processes that are not compliant with the cloud security policies available or you have a state mandate that the data must reside inside of your company walls, then cloud analytics may not fit your use cases. Whatever you decide, make sure you consider all of the aspects to make a well-informed decision.
Analytics On-premise versus in the Cloud
With the current state of public and private cloud offerings, companies have already leaped deploying critical applications to the cloud. Everyday use cases for deploying to the cloud include:
- Application and desktop delivery
- Backup and disaster recovery
- Big data and business analytics
Let’s focus on the last bullet, specifically the deployment of analytics to the cloud. Techopedia defines cloud analytics as “…a type of cloud service model where data analysis and related services are performed on a public or private cloud…”
Before you consider deploying analytics to the cloud, learning a few key marketing phrases that revolve around analytics in the cloud is helpful:
- Cloud analytics
- Big data and business analytics
- Cloud computing
Regardless of what you call it, let’s focus on an important point:
AnalyticsCloud = AnalyticsOn-premise
That’s right. Analytics deployed to the cloud is equal to analytics deployed on-premise. Whether you deploy analytics on-premise or in the cloud, the algorithms and results will be the same. Not just the same – exactly equal! There is no magic happening in the cloud in the world of analytics that did not already happen via an on-premise model. You are not missing some special sauce if you continue to execute your analytic processes outside of the cloud.
The considerations for deploying analytics in the cloud don’t pertain to the results. They relate to the process, deployment, and execution of the analytics. For cloud analytic deployment, the considerations to keep in mind revolve around the process, cost, and infrastructure.
Migration Process – Have You Thought of Everything?
If you are looking at migrating some or all of your organization’s analytics to the cloud, you need to consider the change management involved in making the shift. As with any change management, process changes are always at the top of the list. Process changes to consider include, but are not limited to, the following items:
- Platform management
On the topic of data, make sure that you understand that you need to move the right data to the right place at the right time for the cloud infrastructure to process and return results to the users. This is not a difficult thing to resolve, but you do need to give it some thought before migrating to the cloud. Data does not just show up where you want it to be without some planning in this area.
You need to consider the security implications of moving the data from your local safe haven to the cloud. You are going to be moving your competitive advantage to the cloud. That requires making sure that your position stays safe. The current array of free and commercial SSH and SFTP clients available do an excellent job of providing the necessary security as you move your data across network without worrying about the data or your ideas being hacked.
Performance is something that needs to be worked through. Make sure to work with your user base to adequately measure and test any performance scenarios that you are used to from your non-cloud environment. You want to ensure that your user community does not incur a performance hit from their experience before the shift to the cloud. If you are willing to do the cost-benefit analysis and take your time with some well-designed testing of your processes, performance is something that you can pay to scale so that you are not negatively impacted in this area.
Platform management – managing and orchestrating the cloud platform to your needs – is not a slam dunk in the cloud. It requires planning and coordination to ensure that it will meet your needs. Orchestration software and containers, such as Kubernetes and Docker, are built to help with platform management.
While the above items can be a lot to consider when looking at this paradigm shift, don’t let them scare you away before doing your research. Most cloud software providers can handle all of the above points. Where they can’t, there are quality open source and commercial software solutions to fill in the gap. The methods for how they handle them and how that fits into your corporate processes will help to narrow down who the potential vendors are for your needs.
Cost – Will You Pay More or Less?
When you measure the cost for migrating analytics to the cloud, consider the total cost of ownership (TCO). Make sure to include the following items in your analysis:
- Platform infrastructure
- Technical support
When it comes to looking at cost, the first thing that you need to look at is the cost of the cloud provider. Look at a few different options for cloud providers so you can get comfortable with how they price their infrastructure and the licensing model that they use for one-time vs. recurring costs. Research how often they update their technology. Look at the differences in cost for maintaining a persistent IP address vs. a rotating IP address. If your business does not need 24/7 uptime of the hardware, you will want to look at whether the provider has a model to allow for paying only for the time that you need the servers. If you follow this path, you need to consider how to store and maintain the latest snapshot of your environment so you can pick up from where you left off the next time that you run your processes and create the results.
Cost of the infrastructure does not stop with the hardware. Be sure to plan and know what software is provided for you from the cloud provider so that you can prepare for the procurement of the software that you need to put on that hardware. While the operating system is almost always included in the cost analysis from the cloud provider, analytic software, as well as platform orchestration and management software, need to be considered if they are not available or included by the provider. You need to know whether the software licensing model changes when you go from on-premise to the cloud. You may have been paying by the CPU on-premise, but the cloud licensing model may be by the number of API calls that you make and to which of the algorithms. Alternatively, the cloud licensing model may be on a per-user basis.
When looking at the cost of migrating to the cloud, you need to look at where you will save on human resource costs or where you will have to incur additional headcount as part of your new processes. The odds are outstanding that you will be able to realize cost savings to balance out some of the costs that you will need to incur.
The technical support cost and model is something that you will want to make sure that you understand. If support costs extra for one vendor but not for another, make sure to look at total costs when comparing. If one vendor charges on a per-user basis and another vendor charges for some finite number of users, convert the cost to a per-user calculation to make for an accurate comparison. If the cost of support is included in the price of the software, make sure to check on whether there is a cost for recurring maintenance.
If you are new to the cloud, you will probably need to consider some services to help you to migrate to the cloud. It may only be to help to architect the move. It may be to plan the data strategy. It may be both and more. Make sure that you look at service providers who have performed cloud migration and modernization services previously. This is new for you … you don’t want it to be new for the team who is doing the heavy lifting.
When considering the people costs and savings, don’t forget to include training costs in the analysis. Subtract out any ongoing training costs for your current platform that you will no longer need. Add in the training costs for the new platform that you have identified.
It was mentioned a few times in this section, but do not forget to include the cost savings that you will receive from the above items as you look at the return on investment for the move to the cloud.
Infrastructure Availability and Support
Lastly, make sure that you give attention to infrastructure availability and support considerations. Make sure to consider the following:
- Technical support
When analyzing the availability of the platform, make sure to look at what your users need. You may have gotten so used to having a dedicated server on premise that you did not even consider what percentage of the time that you are using the server. You may only need it for 25% of the day. After reviewing the amount of time that you need the hardware to be available in a given day, make sure to look at the startup time that it takes to fire up the hardware. Make sure that this is an amount of time that is acceptable and include it in your process planning. Gauge whether you need dedicated storage or whether can you work with a storage area network (SAN). Look also at your analytic software’s reliance on I/O channels, networking bandwidth and RAM to make sure that you have the right hardware available at the right times operating at the capacity that you need it to perform.
Be sure to look at the technical support structure of both the hardware and software that you will be leveraging in the cloud. Be sure that their schedule of communication-based on the ticket’s severity level matches your expectation of that communication. Most cloud providers offer 24/7 support. Look at the various channels of support that they offer – email, instant message chats, live person call center. Make sure to look at the SLA support ticket turnaround time and communication. Analytic applications are not always considered operational or mission critical. However, if it is a competitive advantage that your company leverages, it is mission critical to you. The lack of being in the same classification may surprise you in terms of the SLA offered and severity of the support ticket classification that the vendor may employ.
Should I Deploy to the Cloud?
When you have done a thorough job of researching the above items, at a minimum, and you are comfortable with the results, you will be able to make an informed decision on whether cloud analytics is something that you want to pursue.
Remember that you don’t have to (and probably shouldn’t) do the migration all at once. As with any technology implementation or technology shift, success is best guaranteed when you break it down into a few iterative phases rather than one big bang.
Regardless of your decision, make sure that you don’t compromise on quality, cost, performance or timing of the resulting information that you and your organization have already built into a critical competitive advantage.