The cloud offered a completely flexible IT infrastructure, where workloads could be placed exactly where they would provide the best performance, for the best cost, at any time. However, in order for this to happen, businesses need to know exactly what the best option is at any given moment. The complexity required to calculate this is too much to do at a human scale, and even most automated scripts will struggle. In turn, businesses have had to resign themselves to making “best guess” decisions, and unintentionally inviting costly mistakes.
What is the difference between a “hybrid cloud” and what organisations are doing now – aren’t they already using hybrid?
A true hybrid cloud doesn’t just mean using both public cloud and on-premise. It also means that the entire cloud acts as a single environment. Currently, most organisations are using multi-cloud: they have multiple public cloud and on-premises infrastructures, but workloads placed in those infrastructures remain static. While those workloads can be scaled up and down, the organisations will never see the real benefits to performance and cost that a true hybrid cloud will provide until they can take that next step and move workloads between environments as and when necessary. Turbonomic is designed to allow just that.
Monitoring and automating workloads isn’t anything new: why aren’t the currently used approaches good enough?
Simply monitoring or automating workloads isn’t enough. Even if a workload is monitored 24/7, the organisation still needs to be able to make changes in time if the monitoring identifies an issue or an opportunity to improve performance. Conversely, automation alone needs intelligence behind it to ensure it is always making the best decisions, and not potentially harming the business. Turbonomic not only unifies monitoring and automation, but combines workloads, compute, storage and other factors into a market of buyers and sellers of resources, meaning it has the underlying intelligence to make the right decision at the right time.
Turbonomic has claimed it can lower monthly cloud bills by 30% - how does this work, and can you give an example?
Put simply, Turbonomic will lower cloud bills by ensuring that workloads are always placed in the infrastructure that gives them the right performance for the right cost. For instance, an organisation might be running its CRM workload in a public cloud costing $200 per server per month. Turbonomic can identify a similar cloud that can support the same workload and meets the business’s compliance needs, but only costs $180 per server per month, and move the workload there automatically. More importantly, it can be doing this constantly, with every workload, meaning the business is never paying more than it should or allowing performance to suffer.
Isn’t there a risk of running into compliance issues when you automate cloud provision? How do you avoid those?
When automating cloud provision, there are two main compliance issues. First, there is data protection – organisations need to be certain that workloads won’t be placed in infrastructure hosted in regions that don’t meet regulatory demands. Second, there is the issue of software licensing – if a business provisions application workloads that take it beyond its agreed contract for software licenses, it could get a costly shock at its next software audit. To avoid this, these factors need to be included in the intelligence that underlies any automated decision. For instance, in Turbonomic we include data sovereignty and licensing as factors in the market the platform bases its decisions on.
You say that Turbonomic allows businesses to create a “desired state” for workloads: what does that look like, in practice?
The “desired state” for workloads is one where applications have the exact resources they need to guarantee performance, without breaking compliance or incurring unnecessary costs. Quite simply, if an organisation is being as efficient as possible with its resources but performance is suffering then it isn’t in a desired state. Similarly, if all of its applications perform perfectly but it is over-spending on resources then a desired state is way away. Ultimately, a desired state is impossible to precisely define since it’s always changing – our aim with Turbonomic is to help businesses ensure they constantly have that balance.
What are your future plans for Turbonomic – e.g. do you intend to expand its use to Google or other clouds?
We already work with SoftLayer today, and extending to other cloud providers like the Google cloud offering is definitely on the road map. At the moment, the biggest customer demand is for AWS and Azure. It is inevitable that GCP will be getting more enterprise adoption soon which will open the option up for our customers more. Oracle is driving to get on the cloud podium as well, so we are keenly watching the ecosystem to make sure we are working with customers as they adopt these into their hybrid cloud portfolio.