Enterprise IT departments are struggling to deal with a range of options for their enterprise computing platforms. They are faced with a wide set of choices but not all are under their control. The key discussion point seems to be on-premises, cloudor a combination of both, i.e hybrid.
But what cloud option should they choose? They have the choice of private cloud (on-premises), private hosted cloud (using shared infrastructure), public cloud and hybrid cloud. For many companies there is no single answer. It all depends on the application, services, delivery model, how the application will be consumed and, of course, the dreaded regulators with their compliance legislation.
Picking a route through all of this is complicated. The best answer that is beginning to emerge is one of a multicloud environment. What that means is that some elements may stay on-premises, such as data or even some critical services. Other elements may be deployed locally to where they are consumed to get the best possible performance. There is also a more fluid approach where applications and services are moved across multiple clouds based on a number of criteria.
To further confuse the situation, this is only about the cloud elements that IT knows about and has access to. Alongside this, the industry is seeing a steady move by business units to buy access to their own cloud services and applications. They then move corporate data into those services, taking it away from IT.
Virtustream, Forrester and a multicloud research project
To try and understand how all of this is playing out, Virtustream and Forrester set out to look at the state of multicloud in Europe. They talked to a large number of organisations about cloud adoption and the apps that are being migrated onto cloud solutions.
The research was not just about deployment platforms. It looked at how the decision to use cloud matched with both the business objectives and user expectations. Part of this meant taking into account what metrics were being used to assess success or failure.
The result was a complex set of data.
What are companies looking to use cloud for?
The focus of this study was companies that were looking at private, public and multicloud provisioning. In terms of technologies this meant Infrastructure as a Service (IaaS) and Platform as a Service (PaaS). The two options allow companies to both migrate existing apps and build new apps in the cloud. It specifically excluded using SaaS as an app replacement technology.
To get what they wanted, 86% of respondents were targeting multiple cloud platforms. Some were choosing to deploy the same app on multiple clouds based on local availability and performance. Others were making a cost-based judgement.
Migrating cloud workloads based on what it costs to use them can be complicated. At present, it is possible to deploy the same application to multiple cloud platforms using a single deployment script. Keeping that application platform agnostic is harder. Migrating between clouds is harder still as dependencies quickly develop as operations teams tune and configure the application.
The exception to this is the use of virtual machines where they are portable across multiple environments. That market is expanding rapidly with the introduction of containers running microservices. However, the study did not dig deeper into this.
What type of cloud and how is it connected?
The numbers that are touted around the industry at conferences are often highly confusing especially when it comes to what type of ‘cloud’ companies are using. In this survey the majority of those interviewed are using some form of private cloud. This could be private cloud on-premises (66%) or a hosted private cloud (73%). The latter uses some shared infrastructure but is essentially single tenant. Only 56% were using public cloud.
The hosted private cloud and public cloud options also include specialist clouds. This is one of the ways that organisations are finding themselves drawn into a multicloud world. They are partitioning workloads up to the different platforms and then interconnecting them. This raises questions over cost and security.
Building the integration between the clouds is not necessarily difficult. Software providers are delivering APIs to allow this type of coding. Whether the APIs are efficient and easily updated as software changes is another matter. This is where organisations need to review their software design processes.
Securing and moving data
Data security is a top level item for everyone at the moment. Data needs to be encrypted when at rest and, just as importantly, when in transit. It is important that companies make sure they are protecting data.
Performance is also an important factor here. While a specialist cloud might be very good for one application, moving data to/from it, takes time. It may be that companies end up deploying copies of applications in multiple locations and having synchronised data sets that the applications draw upon. For architects, this is a nightmare.
What is this costing?
All of this impacts cost. 44% of respondents are spending at least $50 million on cloud platforms today. A very smallgroup are spending over $200 million per year. These figures are likely to increase as organisations move more workloads to the cloud. This is not just about how much is being moved but the way that cloud seems to increase usage. The underlying computing model allows people to do things that could not be done before. This results in unexpected increases in computing resources which is paid for as part of the cloud cost.
What applications are being moved?
There is a significant focus on mission critical applications in this survey. However, rather than define them as ERP, Accounting, CRM or any other recognised use, the survey uses three other phrases:
- Apps core to completing sales
- Apps core to customer experience
- Apps core to employee productivity
What does that mean? Instead of ERP it could simply mean Excel and Word when to comes to creating quotes and sending out invoices. Maybe good customer experience is about the latest chatbot or perhaps an FAQ written in simple language. As to employee productivity that’s anyone’s guess. However, these are not some arbitrary Virtustream or Forrester terms but apparently how the C-Suite and senior management describe critical apps.
What is important is that EMEA is lagging behind the rest of the world when it comes to moving these to the cloud. Only 57% of companies in EMEA are moving these to the cloud compared to 60% in the US and 65% in APAC. It’s not entirely clear why this should be but it does mirror cloud adoption and other technology take-up rates.
Europe has always been slower to adopt technology trends than the other two regions. As GDPR kicks in, this could slow that take-up even more, at least in terms of public cloud. This may also figure in the decision of 41% of companies in EMEA who said they were not planning to move mission-critical apps to the cloud.
What are the reasons for using cloud?
The early cloud hype was all about dumping costly infrastructure, slash CAPEX and move to an OPEX model. That has changed as reality set in and companies realised that OPEX had the potential to quickly spiral out of control.
The respondents are now focused on increasing operational efficiency, improving products and service, and improving their ability to innovate. Lower cost comes after these and, surprisingly, ahead of growing revenue. This is an important shift. It shows that IT is transitioning more and more to a service organisation. Importantly, it seems that this also aligns with the organisations business objectives.
While cloud is key to those objectives it is interesting to note that there are still issues that cloud is not addressing. Lowering IT costs is anissue that will need to be looked at over time, especially in light of who runs what workload. SaaS companies are able to provide access to functionality that most organisations would call mission-critical.
This can be seen by many of the cloud-based ERP and CRM platforms. They deliver all the underlying code and instead of spending large numbers of man hours and money customising applications, customers just configure them. This shift allows for new functionality to be added quickly rather than the traditional coding cycles. It aligns with those operational goals of meeting the expectations of customers, both internal and external. But, it requires organisations to manage an initial complex migration that is often costly, for the promise of a better life in the future.
Different regions want different things from multicloud
The expectations from multicloud vary by region but there is significant overlap in both expectation and challenges. The use of specialty clouds to match the differing requirements of workloads is a key driver here. It should be no surprise. It matches the way hardware was purchased and configured inside the datacentre based on workload.
Using specific clouds to deal with security and compliance is on the agenda. This is increasingly important as privacy and geo-locking of data become more commonplace. Choosing cloud services based on cost would also appear to make sense. As IT continues to look for savings, including how to charge back computing costs to business units, it will have to show it is being effective with its spending. However, the caveat here is that it has to prove that taking the lowest cost option does not increase risk or lower service levels.
There is also an expectation, yet to be realised, that multicloud will also deliver improved management. Organisations are struggling to integrate their on-premises and private clouds with public clouds. The more clouds that they add in, the greater the problem can become. This is not a new issue nor is it one that is easily solved. There are open source, multi-platform management suites. What they gain in support for platforms is often lost in terms of functionality.
Security across multicloud is also seen as challenging. In reality, it shouldn’t be. Data is either encrypted or not encrypted when stored on the cloud. There is no reason why any corporate would take on a solution where disk encryption wasn’t a core offering. The same is true for data in transit. Where this gets tricky is for those enterprises that want to bring their own encryption solutions and keys. It requires the platforms to support this and that is more complicated than people realise.
Although SaaS was not part of this survey, it is a bigger issue in terms of end-user cloud usage. The majority of users and even business units will buy services based on features. They do not know and are not able to determine the difference between consumer and business level security. This means that a lot of data is already at risk. Bringing that into a homogenised security model will take time.
One of the biggest reasons for multicloud is avoiding vendor lock in. This has always been an issue and is still going to be an issue with cloud-based applications. It is a non-trivial task to port a database and an even bigger challenge when the data is used in ERP, CRM or accounting software. When it comes to migrating workloads, there is much to be done to prove that VMs and containers are the ideal solution. Other approaches are complex and will often cost more than negotiating a new deal on the existing platform.
Cloud, in all its variations, has become a valid deployment platform. Enterprise IT is continuing to adopt it and move workloads away its data centres. For all the current successes, however, there are still many challenges to be resolved.
What is important is that the vendors are engaging with enterprise IT which is engaging with its users. The board is setting operational targets and goals that IT is able to meet with the current generation of cloud. It will be interesting to see how long it takes to get to the goal of a commoditised multicloud world.