InsightaaS: Wikibon has released an interesting video discussing several different approaches to hybrid computing – one which helps clarify both deployment options and the positioning of various suppliers.
The video begins with Wikibon analyst Brian Gracely observing that “hybrid cloud is not a physical thing,” which means that it can be defined and deployed in different ways. Gracely and Wikibon’s Stuart Miniman then describe a number of different deployment approaches, including references to vendors aligned with the approach:
- A mix of on-premise hardware with integrated software, and the same configuration in the public cloud. This has the advantage of offering consistent, predictable performance.
- Supporting vendors: The Wikibon team notes that Microsoft has been making announcements in this area (with the Dell Stack and HP stack), and that Oracle Exadata fits this model as well.
- Customer operates on-premise hardware, but the interconnection is use of the same software both on-premise and in the cloud. This increases flexibility, but may encounter issues connected to different performance in different locations, and hardware update timing can be an issue as well.
- Supporting vendors: Microsoft via Microsoft Server/Azure; VMware with vCenter/vCloud Air.
- API-level integration, interoperability at the API level (storage, network, etc.). The video uses an example configuration with Red Hat on premise and Rackspace in the cloud to illustrate this approach, which they believe offers “lots of openness and flexibility,” but at the cost of interoperability that is not as seamless as might be desired – Wikibon believes that there are limitations in interoperability across distributions, with Miniman stating that “portability is something that OpenStack hasn’t yet delivered on.”
- Hybrid that starts off-premise, adding an on-premise instance to enable local data storage. Platform9, ZeroStack and IBM’s Blue Box are offered as examples of vendors offerings in this space.
- “Broker model,” where a customer can run VMware, OpenStack, Docker or similar technologies, and treat multiple public clouds as resources. Wikibon sees VCE pointing here, but notes that today, this is a complex proposition.
- Beyond infrastructure – management/integration of SaaS clouds. Wikibon data shows that two-thirds of public cloud is SaaS. How, Gracely and Miniman wonder, will enterprises manage these resources? Will they be IT-managed or business-managed? Will they need to be integrated with other resources?
The video ends with two observations. One is that hybrid cloud is evolving. The second is that “the more consistent you are, the more possible it is to do cloud today; the more variability you want in [your cloud approach]. The more variability you want, the more clouds you want to use, the more applications – the more complex [a cloud scenario] is, the less mature that is today.” Complexity isn’t itself a reason not to pursue a hybrid strategy, though: as Miniman noted, “users today are looking for more sources of data,” and users who rely primarily or entirely on on-premise infrastructure will find that “getting to other data sources is tough,” while those whose operations are house in “one of those giant environments that has 20 different clouds” will have much greater access to external communities of data.
The video is 11 minutes long, and a very worthwhile use of those minutes!
