November 23, 2016 | Written by: Andras Szakal
This post continues the discussion on the importance of the hybrid cloud datacenter and how to approach migration. In reflecting back on part one of this discussion, it was evident that the speed of innovation and the rate at which these platforms are evolving in and of itself must be considered as part of your cloud migration journey.
The Datacenter as a Hybrid Cloud Broker
Large highly transactional systems might not lend themselves to virtual public instances – the exception might be a bare metal environment. They may well not even fit into some cloud provider virtual appliances or VM single instances. Over the last year, I have helped several agencies that are considering moving both mission and business applications. These applications have several characteristics in common – they are highly transactional coupled with large in-memory database or application foot prints. Many of the public cloud virtual appliance based cloud instances can only sustain a maximum of ~40 core 256 GB virtual environment. Large financial and mission systems – what I would call core workloads – tend to bump up against the maximum capacity of many public cloud environments, especially if these applications were not born in the cloud. These systems contain and manage your most important organizational data and are often referred to as Systems of Record.
Furthermore, core commercial cloud provider infrastructure is intended for servicing the nominal workload and therefore highly optimized enterprise environments are not really supported today – for example, large database and transactional systems often require these optimizations, such as database solutions that require fiber channel connections between clustered systems. Another example is an organization that uses Parallel Sysplex technology to linearly scale while concurrently accessing shared data. You will not be able to replicate those systems directly in the commercial cloud. None of the commercial cloud providers can support these requirements using their nominal commercial offerings – nor should they try. As such, most all of the CSPs now have some partnership or solution for integrating large hosted transactional or enterprise systems outside the CSP environment but directly connected to the CSP. This allows integration and access to system APIs that are necessary to build new cloud based functionality. Of course, this requires extending those core systems via APIs so they can be reused by new, born-on-the-cloud applications.
Traditional enterprise datacenters are transforming into hybrid cloud brokers. The new hybrid cloud datacenter must provide support into the cloud for those core systems with their high value data needs to be accessible by cloud based applications and in order to build new applications born in the cloud and enabled for multimodal mobile access. This requires the application of new and evolving hybrid cloud management solutions that can help manage access to multiple cloud providers and facilitate the extension of core systems into the cloud while simultaneously protecting core systems or data.
We refer to this integration with external cloud services as the economy of APIs. And API is an important distinction – because it’s the APIs that define these services and the coordination of access to these APIs including down to which version of the API that is an essential responsibility of the future datacenter.
Of course security is an essential service of the hybrid cloud data center as the connectivity nexus to the entire external commercial or partner cloud based environment. Of paramount importance is the ability to federate a standard IDaaS and directory service across cloud platforms. If your cloud provider is not thinking about how to address federated identify management then you will have difficulty managing the processes (APIs) that comprise composite services.
Extending the datacenter into the cloud
And lastly, the datacenter – the hybrid nexus – that glues the cloud cohesively will become the cloud broker to ensure that the enterprise can effectively realize the proper ROI from multiple cloud providers using an approach that is organized and predictable in what that gives you choice with consistency. Whether it’s bursting out to support unpredictable loads or aggregating sensor information from IOT solutions — you will not be managing just one cloud provider, but multiple – the future enterprise datacenter must be the center of gravity for managing those interactions.
Of course your datacenter must embrace the same underlying infrastructure and development technologies as the commercial cloud provider. This means implementing a provisioning and orchestration backplane that enables your staff to automate the management and deployment of datacenter infrastructure – compute, storage and network: all have to become software defined – with the goal of enabling and facilitating the integration between multiple cloud providers and many cloud enabled services. This is where new hybrid cloud management solutions come into play.
Ultimately, your development approach must eventually support a two-speed but agile enterprise development and operations approach. This means implementing DevOps practices and tools that allow for effective integration with cloud services while effectively managing a more structured rate and pace of change to the core systems of record. Eventually, enterprises will need to migrate and refactor older brittle monolithic enterprise systems using a more progressive cloud-based architecture. Otherwise, your organization and datacenter will suffer from an impedance mismatch manifested when older second-generation architecture is unable to keep pace with almost frictionless born-in-the-cloud development and deployment.